1 <?xml version="1.0" encoding="iso-8859-1"?>
2 <sect1 id="runtime-control">
3 <title>Running a compiled program</title>
5 <indexterm><primary>runtime control of Haskell programs</primary></indexterm>
6 <indexterm><primary>running, compiled program</primary></indexterm>
7 <indexterm><primary>RTS options</primary></indexterm>
9 <para>To make an executable program, the GHC system compiles your
10 code and then links it with a non-trivial runtime system (RTS),
11 which handles storage management, profiling, etc.</para>
13 <para>You have some control over the behaviour of the RTS, by giving
14 special command-line arguments to your program.</para>
16 <para>When your Haskell program starts up, its RTS extracts
17 command-line arguments bracketed between
18 <option>+RTS</option><indexterm><primary><option>+RTS</option></primary></indexterm>
20 <option>-RTS</option><indexterm><primary><option>-RTS</option></primary></indexterm>
21 as its own. For example:</para>
24 % ./a.out -f +RTS -p -S -RTS -h foo bar
27 <para>The RTS will snaffle <option>-p</option> <option>-S</option>
28 for itself, and the remaining arguments <literal>-f -h foo bar</literal>
29 will be handed to your program if/when it calls
30 <function>System.getArgs</function>.</para>
32 <para>No <option>-RTS</option> option is required if the
33 runtime-system options extend to the end of the command line, as in
37 % hls -ltr /usr/etc +RTS -A5m
40 <para>If you absolutely positively want all the rest of the options
41 in a command line to go to the program (and not the RTS), use a
42 <option>––RTS</option><indexterm><primary><option>--RTS</option></primary></indexterm>.</para>
44 <para>As always, for RTS options that take
45 <replaceable>size</replaceable>s: If the last character of
46 <replaceable>size</replaceable> is a K or k, multiply by 1000; if an
47 M or m, by 1,000,000; if a G or G, by 1,000,000,000. (And any
48 wraparound in the counters is <emphasis>your</emphasis>
51 <para>Giving a <literal>+RTS -f</literal>
52 <indexterm><primary><option>-f</option></primary><secondary>RTS option</secondary></indexterm> option
53 will print out the RTS options actually available in your program
54 (which vary, depending on how you compiled).</para>
56 <para>NOTE: since GHC is itself compiled by GHC, you can change RTS
57 options in the compiler using the normal
58 <literal>+RTS ... -RTS</literal>
59 combination. eg. to increase the maximum heap
60 size for a compilation to 128M, you would add
61 <literal>+RTS -M128m -RTS</literal>
62 to the command line.</para>
64 <sect2 id="rts-optinos-environment">
65 <title>Setting global RTS options</title>
67 <indexterm><primary>RTS options</primary><secondary>from the environment</secondary></indexterm>
68 <indexterm><primary>environment variable</primary><secondary>for
69 setting RTS options</secondary></indexterm>
71 <para>RTS options are also taken from the environment variable
72 <envar>GHCRTS</envar><indexterm><primary><envar>GHCRTS</envar></primary>
73 </indexterm>. For example, to set the maximum heap size
74 to 128M for all GHC-compiled programs (using an
75 <literal>sh</literal>-like shell):</para>
82 <para>RTS options taken from the <envar>GHCRTS</envar> environment
83 variable can be overridden by options given on the command
88 <sect2 id="rts-options-misc">
89 <title>Miscellaneous RTS options</title>
93 <term><option>-V<replaceable>secs</replaceable></option>
94 <indexterm><primary><option>-V</option></primary><secondary>RTS
95 option</secondary></indexterm></term>
97 <para>Sets the interval that the RTS clock ticks at. The
98 runtime uses a single timer signal to count ticks; this timer
99 signal is used to control the context switch timer (<xref
100 linkend="using-concurrent" />) and the heap profiling
101 timer <xref linkend="rts-options-heap-prof" />. Also, the
102 time profiler uses the RTS timer signal directly to record
103 time profiling samples.</para>
105 <para>Normally, setting the <option>-V</option> option
106 directly is not necessary: the resolution of the RTS timer is
107 adjusted automatically if a short interval is requested with
108 the <option>-C</option> or <option>-i</option> options.
109 However, setting <option>-V</option> is required in order to
110 increase the resolution of the time profiler.</para>
112 <para>Using a value of zero disables the RTS clock
113 completely, and has the effect of disabling timers that
114 depend on it: the context switch timer and the heap profiling
115 timer. Context switches will still happen, but
116 deterministically and at a rate much faster than normal.
117 Disabling the interval timer is useful for debugging, because
118 it eliminates a source of non-determinism at runtime.</para>
123 <term><option>--install-signal-handlers=<replaceable>yes|no</replaceable></option>
124 <indexterm><primary><option>--install-signal-handlers</option></primary><secondary>RTS
125 option</secondary></indexterm></term>
127 <para>If yes (the default), the RTS installs signal handlers to catch
128 things like ctrl-C. This option is primarily useful for when
129 you are using the Haskell code as a DLL, and want to set your
130 own signal handlers.</para>
136 <sect2 id="rts-options-gc">
137 <title>RTS options to control the garbage collector</title>
139 <indexterm><primary>garbage collector</primary><secondary>options</secondary></indexterm>
140 <indexterm><primary>RTS options</primary><secondary>garbage collection</secondary></indexterm>
142 <para>There are several options to give you precise control over
143 garbage collection. Hopefully, you won't need any of these in
144 normal operation, but there are several things that can be tweaked
145 for maximum performance.</para>
151 <option>-A</option><replaceable>size</replaceable>
152 <indexterm><primary><option>-A</option></primary><secondary>RTS option</secondary></indexterm>
153 <indexterm><primary>allocation area, size</primary></indexterm>
156 <para>[Default: 256k] Set the allocation area size
157 used by the garbage collector. The allocation area
158 (actually generation 0 step 0) is fixed and is never resized
159 (unless you use <option>-H</option>, below).</para>
161 <para>Increasing the allocation area size may or may not
162 give better performance (a bigger allocation area means
163 worse cache behaviour but fewer garbage collections and less
166 <para>With only 1 generation (<option>-G1</option>) the
167 <option>-A</option> option specifies the minimum allocation
168 area, since the actual size of the allocation area will be
169 resized according to the amount of data in the heap (see
170 <option>-F</option>, below).</para>
177 <indexterm><primary><option>-c</option></primary><secondary>RTS option</secondary></indexterm>
178 <indexterm><primary>garbage collection</primary><secondary>compacting</secondary></indexterm>
179 <indexterm><primary>compacting garbage collection</primary></indexterm>
182 <para>Use a compacting algorithm for collecting the oldest
183 generation. By default, the oldest generation is collected
184 using a copying algorithm; this option causes it to be
185 compacted in-place instead. The compaction algorithm is
186 slower than the copying algorithm, but the savings in memory
187 use can be considerable.</para>
189 <para>For a given heap size (using the <option>-H</option>
190 option), compaction can in fact reduce the GC cost by
191 allowing fewer GCs to be performed. This is more likely
192 when the ratio of live data to heap size is high, say
193 >30%.</para>
195 <para>NOTE: compaction doesn't currently work when a single
196 generation is requested using the <option>-G1</option>
202 <term><option>-c</option><replaceable>n</replaceable></term>
205 <para>[Default: 30] Automatically enable
206 compacting collection when the live data exceeds
207 <replaceable>n</replaceable>% of the maximum heap size
208 (see the <option>-M</option> option). Note that the maximum
209 heap size is unlimited by default, so this option has no
210 effect unless the maximum heap size is set with
211 <option>-M</option><replaceable>size</replaceable>. </para>
217 <option>-F</option><replaceable>factor</replaceable>
218 <indexterm><primary><option>-F</option></primary><secondary>RTS option</secondary></indexterm>
219 <indexterm><primary>heap size, factor</primary></indexterm>
223 <para>[Default: 2] This option controls the amount
224 of memory reserved for the older generations (and in the
225 case of a two space collector the size of the allocation
226 area) as a factor of the amount of live data. For example,
227 if there was 2M of live data in the oldest generation when
228 we last collected it, then by default we'll wait until it
229 grows to 4M before collecting it again.</para>
231 <para>The default seems to work well here. If you have
232 plenty of memory, it is usually better to use
233 <option>-H</option><replaceable>size</replaceable> than to
235 <option>-F</option><replaceable>factor</replaceable>.</para>
237 <para>The <option>-F</option> setting will be automatically
238 reduced by the garbage collector when the maximum heap size
239 (the <option>-M</option><replaceable>size</replaceable>
240 setting) is approaching.</para>
246 <option>-G</option><replaceable>generations</replaceable>
247 <indexterm><primary><option>-G</option></primary><secondary>RTS option</secondary></indexterm>
248 <indexterm><primary>generations, number of</primary></indexterm>
251 <para>[Default: 2] Set the number of generations
252 used by the garbage collector. The default of 2 seems to be
253 good, but the garbage collector can support any number of
254 generations. Anything larger than about 4 is probably not a
255 good idea unless your program runs for a
256 <emphasis>long</emphasis> time, because the oldest
257 generation will hardly ever get collected.</para>
259 <para>Specifying 1 generation with <option>+RTS -G1</option>
260 gives you a simple 2-space collector, as you would expect.
261 In a 2-space collector, the <option>-A</option> option (see
262 above) specifies the <emphasis>minimum</emphasis> allocation
263 area size, since the allocation area will grow with the
264 amount of live data in the heap. In a multi-generational
265 collector the allocation area is a fixed size (unless you
266 use the <option>-H</option> option, see below).</para>
272 <option>-g</option><replaceable>threads</replaceable>
273 <indexterm><primary><option>-g</option></primary><secondary>RTS option</secondary></indexterm>
276 <para>[Default: 1] [new in GHC 6.10] Set the number
277 of threads to use for garbage collection. This option is
278 only accepted when the program was linked with the
279 <option>-threaded</option> option; see <xref
280 linkend="options-linker" />.</para>
282 <para>The garbage collector is able to work in parallel when
283 given more than one OS thread. Experiments have shown
284 that this usually results in a performance improvement
285 given 3 cores or more; with 2 cores it may or may not be
286 beneficial, depending on the workload. Bigger heaps work
287 better with parallel GC, so set your <option>-H</option>
288 value high (3 or more times the maximum residency). Look
289 at the timing stats with <option>+RTS -s</option> to
290 see whether you're getting any benefit from parallel GC or
291 not. If you find parallel GC is
292 significantly <emphasis>slower</emphasis> (in elapsed
293 time) than sequential GC, please report it as a
296 <para>This value is set automatically when the
297 <option>-N</option> option is used, so the only reason to
298 use <option>-g</option> would be if you wanted to use a
299 different number of threads for GC than for execution.
300 For example, if your program is strictly single-threaded
301 but you still want to benefit from parallel GC, then it
302 might make sense to use <option>-g</option> rather than
303 <option>-N</option>.</para>
309 <option>-H</option><replaceable>size</replaceable>
310 <indexterm><primary><option>-H</option></primary><secondary>RTS option</secondary></indexterm>
311 <indexterm><primary>heap size, suggested</primary></indexterm>
314 <para>[Default: 0] This option provides a
315 “suggested heap size” for the garbage collector. The
316 garbage collector will use about this much memory until the
317 program residency grows and the heap size needs to be
318 expanded to retain reasonable performance.</para>
320 <para>By default, the heap will start small, and grow and
321 shrink as necessary. This can be bad for performance, so if
322 you have plenty of memory it's worthwhile supplying a big
323 <option>-H</option><replaceable>size</replaceable>. For
324 improving GC performance, using
325 <option>-H</option><replaceable>size</replaceable> is
326 usually a better bet than
327 <option>-A</option><replaceable>size</replaceable>.</para>
333 <option>-I</option><replaceable>seconds</replaceable>
334 <indexterm><primary><option>-I</option></primary>
335 <secondary>RTS option</secondary>
337 <indexterm><primary>idle GC</primary>
341 <para>(default: 0.3) In the threaded and SMP versions of the RTS (see
342 <option>-threaded</option>, <xref linkend="options-linker" />), a
343 major GC is automatically performed if the runtime has been idle
344 (no Haskell computation has been running) for a period of time.
345 The amount of idle time which must pass before a GC is performed is
346 set by the <option>-I</option><replaceable>seconds</replaceable>
347 option. Specifying <option>-I0</option> disables the idle GC.</para>
349 <para>For an interactive application, it is probably a good idea to
350 use the idle GC, because this will allow finalizers to run and
351 deadlocked threads to be detected in the idle time when no Haskell
352 computation is happening. Also, it will mean that a GC is less
353 likely to happen when the application is busy, and so
354 responsiveness may be improved. However, if the amount of live data in
355 the heap is particularly large, then the idle GC can cause a
356 significant delay, and too small an interval could adversely affect
357 interactive responsiveness.</para>
359 <para>This is an experimental feature, please let us know if it
360 causes problems and/or could benefit from further tuning.</para>
366 <option>-k</option><replaceable>size</replaceable>
367 <indexterm><primary><option>-k</option></primary><secondary>RTS option</secondary></indexterm>
368 <indexterm><primary>stack, minimum size</primary></indexterm>
371 <para>[Default: 1k] Set the initial stack size for
372 new threads. Thread stacks (including the main thread's
373 stack) live on the heap, and grow as required. The default
374 value is good for concurrent applications with lots of small
375 threads; if your program doesn't fit this model then
376 increasing this option may help performance.</para>
378 <para>The main thread is normally started with a slightly
379 larger heap to cut down on unnecessary stack growth while
380 the program is starting up.</para>
386 <option>-K</option><replaceable>size</replaceable>
387 <indexterm><primary><option>-K</option></primary><secondary>RTS option</secondary></indexterm>
388 <indexterm><primary>stack, maximum size</primary></indexterm>
391 <para>[Default: 8M] Set the maximum stack size for
392 an individual thread to <replaceable>size</replaceable>
393 bytes. This option is there purely to stop the program
394 eating up all the available memory in the machine if it gets
395 into an infinite loop.</para>
401 <option>-m</option><replaceable>n</replaceable>
402 <indexterm><primary><option>-m</option></primary><secondary>RTS option</secondary></indexterm>
403 <indexterm><primary>heap, minimum free</primary></indexterm>
406 <para>Minimum % <replaceable>n</replaceable> of heap
407 which must be available for allocation. The default is
414 <option>-M</option><replaceable>size</replaceable>
415 <indexterm><primary><option>-M</option></primary><secondary>RTS option</secondary></indexterm>
416 <indexterm><primary>heap size, maximum</primary></indexterm>
419 <para>[Default: unlimited] Set the maximum heap size to
420 <replaceable>size</replaceable> bytes. The heap normally
421 grows and shrinks according to the memory requirements of
422 the program. The only reason for having this option is to
423 stop the heap growing without bound and filling up all the
424 available swap space, which at the least will result in the
425 program being summarily killed by the operating
428 <para>The maximum heap size also affects other garbage
429 collection parameters: when the amount of live data in the
430 heap exceeds a certain fraction of the maximum heap size,
431 compacting collection will be automatically enabled for the
432 oldest generation, and the <option>-F</option> parameter
433 will be reduced in order to avoid exceeding the maximum heap
440 <option>-t</option><optional><replaceable>file</replaceable></optional>
441 <indexterm><primary><option>-t</option></primary><secondary>RTS option</secondary></indexterm>
444 <option>-s</option><optional><replaceable>file</replaceable></optional>
445 <indexterm><primary><option>-s</option></primary><secondary>RTS option</secondary></indexterm>
448 <option>-S</option><optional><replaceable>file</replaceable></optional>
449 <indexterm><primary><option>-S</option></primary><secondary>RTS option</secondary></indexterm>
452 <para>These options produce runtime-system statistics, such
453 as the amount of time spent executing the program and in the
454 garbage collector, the amount of memory allocated, the
455 maximum size of the heap, and so on. The three
456 variants give different levels of detail:
457 <option>-t</option> produces a single line of output in the
458 same format as GHC's <option>-Rghc-timing</option> option,
459 <option>-s</option> produces a more detailed summary at the
460 end of the program, and <option>-S</option> additionally
461 produces information about each and every garbage
464 <para>The output is placed in
465 <replaceable>file</replaceable>. If
466 <replaceable>file</replaceable> is omitted, then the output
467 is sent to <constant>stderr</constant>.</para>
475 <title>RTS options for profiling and parallelism</title>
477 <para>The RTS options related to profiling are described in <xref
478 linkend="rts-options-heap-prof"/>, those for concurrency in
479 <xref linkend="using-concurrent" />, and those for parallelism in
480 <xref linkend="parallel-options"/>.</para>
483 <sect2 id="rts-options-debugging">
484 <title>RTS options for hackers, debuggers, and over-interested
487 <indexterm><primary>RTS options, hacking/debugging</primary></indexterm>
489 <para>These RTS options might be used (a) to avoid a GHC bug,
490 (b) to see “what's really happening”, or
491 (c) because you feel like it. Not recommended for everyday
499 <indexterm><primary><option>-B</option></primary><secondary>RTS option</secondary></indexterm>
502 <para>Sound the bell at the start of each (major) garbage
505 <para>Oddly enough, people really do use this option! Our
506 pal in Durham (England), Paul Callaghan, writes: “Some
507 people here use it for a variety of
508 purposes—honestly!—e.g., confirmation that the
509 code/machine is doing something, infinite loop detection,
510 gauging cost of recently added code. Certain people can even
511 tell what stage [the program] is in by the beep
512 pattern. But the major use is for annoying others in the
513 same office…”</para>
519 <option>-D</option><replaceable>num</replaceable>
520 <indexterm><primary>-D</primary><secondary>RTS option</secondary></indexterm>
523 <para>An RTS debugging flag; varying quantities of output
524 depending on which bits are set in
525 <replaceable>num</replaceable>. Only works if the RTS was
526 compiled with the <option>DEBUG</option> option.</para>
532 <option>-r</option><replaceable>file</replaceable>
533 <indexterm><primary><option>-r</option></primary><secondary>RTS option</secondary></indexterm>
534 <indexterm><primary>ticky ticky profiling</primary></indexterm>
535 <indexterm><primary>profiling</primary><secondary>ticky ticky</secondary></indexterm>
538 <para>Produce “ticky-ticky” statistics at the
539 end of the program run. The <replaceable>file</replaceable>
540 business works just like on the <option>-S</option> RTS
541 option (above).</para>
543 <para>“Ticky-ticky” statistics are counts of
544 various program actions (updates, enters, etc.) The program
545 must have been compiled using
546 <option>-ticky</option><indexterm><primary><option>-ticky</option></primary></indexterm>
547 (a.k.a. “ticky-ticky profiling”), and, for it to
548 be really useful, linked with suitable system libraries.
549 Not a trivial undertaking: consult the installation guide on
550 how to set things up for easy “ticky-ticky”
551 profiling. For more information, see <xref
552 linkend="ticky-ticky"/>.</para>
559 <indexterm><primary><option>-xc</option></primary><secondary>RTS option</secondary></indexterm>
562 <para>(Only available when the program is compiled for
563 profiling.) When an exception is raised in the program,
564 this option causes the current cost-centre-stack to be
565 dumped to <literal>stderr</literal>.</para>
567 <para>This can be particularly useful for debugging: if your
568 program is complaining about a <literal>head []</literal>
569 error and you haven't got a clue which bit of code is
570 causing it, compiling with <literal>-prof
571 -auto-all</literal> and running with <literal>+RTS -xc
572 -RTS</literal> will tell you exactly the call stack at the
573 point the error was raised.</para>
575 <para>The output contains one line for each exception raised
576 in the program (the program might raise and catch several
577 exceptions during its execution), where each line is of the
581 < cc<subscript>1</subscript>, ..., cc<subscript>n</subscript> >
583 <para>each <literal>cc</literal><subscript>i</subscript> is
584 a cost centre in the program (see <xref
585 linkend="cost-centres"/>), and the sequence represents the
586 “call stack” at the point the exception was
587 raised. The leftmost item is the innermost function in the
588 call stack, and the rightmost item is the outermost
597 <indexterm><primary><option>-Z</option></primary><secondary>RTS option</secondary></indexterm>
600 <para>Turn <emphasis>off</emphasis> “update-frame
601 squeezing” at garbage-collection time. (There's no
602 particularly good reason to turn it off, except to ensure
603 the accuracy of certain data collected regarding thunk entry
611 <sect2 id="rts-hooks">
612 <title>“Hooks” to change RTS behaviour</title>
614 <indexterm><primary>hooks</primary><secondary>RTS</secondary></indexterm>
615 <indexterm><primary>RTS hooks</primary></indexterm>
616 <indexterm><primary>RTS behaviour, changing</primary></indexterm>
618 <para>GHC lets you exercise rudimentary control over the RTS
619 settings for any given program, by compiling in a
620 “hook” that is called by the run-time system. The RTS
621 contains stub definitions for all these hooks, but by writing your
622 own version and linking it on the GHC command line, you can
623 override the defaults.</para>
625 <para>Owing to the vagaries of DLL linking, these hooks don't work
626 under Windows when the program is built dynamically.</para>
628 <para>The hook <literal>ghc_rts_opts</literal><indexterm><primary><literal>ghc_rts_opts</literal></primary>
629 </indexterm>lets you set RTS
630 options permanently for a given program. A common use for this is
631 to give your program a default heap and/or stack size that is
632 greater than the default. For example, to set <literal>-H128m
633 -K1m</literal>, place the following definition in a C source
637 char *ghc_rts_opts = "-H128m -K1m";
640 <para>Compile the C file, and include the object file on the
641 command line when you link your Haskell program.</para>
643 <para>These flags are interpreted first, before any RTS flags from
644 the <literal>GHCRTS</literal> environment variable and any flags
645 on the command line.</para>
647 <para>You can also change the messages printed when the runtime
648 system “blows up,” e.g., on stack overflow. The hooks
649 for these are as follows:</para>
655 <function>void OutOfHeapHook (unsigned long, unsigned long)</function>
656 <indexterm><primary><function>OutOfHeapHook</function></primary></indexterm>
659 <para>The heap-overflow message.</para>
665 <function>void StackOverflowHook (long int)</function>
666 <indexterm><primary><function>StackOverflowHook</function></primary></indexterm>
669 <para>The stack-overflow message.</para>
675 <function>void MallocFailHook (long int)</function>
676 <indexterm><primary><function>MallocFailHook</function></primary></indexterm>
679 <para>The message printed if <function>malloc</function>
685 <para>For examples of the use of these hooks, see GHC's own
687 <filename>ghc/compiler/parser/hschooks.c</filename> in a GHC
692 <title>Getting information about the RTS</title>
694 <indexterm><primary>RTS</primary></indexterm>
696 <para>It is possible to ask the RTS to give some information about
697 itself. To do this, use the <option>--info</option> flag, e.g.</para>
699 $ ./a.out +RTS --info
701 ,("GHC version", "6.7")
702 ,("RTS way", "rts_p")
703 ,("Host platform", "x86_64-unknown-linux")
704 ,("Build platform", "x86_64-unknown-linux")
705 ,("Target platform", "x86_64-unknown-linux")
706 ,("Compiler unregisterised", "NO")
707 ,("Tables next to code", "YES")
710 <para>The information is formatted such that it can be read as a
711 of type <literal>[(String, String)]</literal>.</para>
716 ;;; Local Variables: ***
718 ;;; sgml-parent-document: ("users_guide.xml" "book" "chapter" "sect1") ***