1 <?xml version="1.0" encoding="iso-8859-1"?>
2 <sect1 id="options-debugging">
3 <title>Debugging the compiler</title>
5 <indexterm><primary>debugging options (for GHC)</primary></indexterm>
7 <para>HACKER TERRITORY. HACKER TERRITORY. (You were warned.)</para>
9 <sect2 id="dumping-output">
10 <title>Dumping out compiler intermediate structures</title>
12 <indexterm><primary>dumping GHC intermediates</primary></indexterm>
13 <indexterm><primary>intermediate passes, output</primary></indexterm>
18 <option>-ddump-</option><replaceable>pass</replaceable>
19 <indexterm><primary><option>-ddump</option> options</primary></indexterm>
22 <para>Make a debugging dump after pass
23 <literal><pass></literal> (may be common enough to need
24 a short form…). You can get all of these at once
25 (<emphasis>lots</emphasis> of output) by using
26 <option>-v5</option>, or most of them with
27 <option>-v4</option>. Some of the most useful ones
33 <option>-ddump-parsed</option>:
34 <indexterm><primary><option>-ddump-parsed</option></primary></indexterm>
37 <para>parser output</para>
43 <option>-ddump-rn</option>:
44 <indexterm><primary><option>-ddump-rn</option></primary></indexterm>
47 <para>renamer output</para>
53 <option>-ddump-tc</option>:
54 <indexterm><primary><option>-ddump-tc</option></primary></indexterm>
57 <para>typechecker output</para>
63 <option>-ddump-splices</option>:
64 <indexterm><primary><option>-ddump-splices</option></primary></indexterm>
67 <para>Dump Template Haskell expressions that we splice in,
68 and what Haskell code the expression evaluates to.</para>
74 <option>-ddump-types</option>:
75 <indexterm><primary><option>-ddump-types</option></primary></indexterm>
78 <para>Dump a type signature for each value defined at
79 the top level of the module. The list is sorted
80 alphabetically. Using <option>-dppr-debug</option>
81 dumps a type signature for all the imported and
82 system-defined things as well; useful for debugging the
89 <option>-ddump-deriv</option>:
90 <indexterm><primary><option>-ddump-deriv</option></primary></indexterm>
93 <para>derived instances</para>
99 <option>-ddump-ds</option>:
100 <indexterm><primary><option>-ddump-ds</option></primary></indexterm>
103 <para>desugarer output</para>
109 <option>-ddump-spec</option>:
110 <indexterm><primary><option>-ddump-spec</option></primary></indexterm>
113 <para>output of specialisation pass</para>
119 <option>-ddump-rules</option>:
120 <indexterm><primary><option>-ddump-rules</option></primary></indexterm>
123 <para>dumps all rewrite rules specified in this module;
124 see <xref linkend="controlling-rules"/>.
131 <option>-ddump-simpl</option>:
132 <indexterm><primary><option>-ddump-simpl</option></primary></indexterm>
135 <para>simplifier output (Core-to-Core passes)</para>
141 <option>-ddump-inlinings</option>:
142 <indexterm><primary><option>-ddump-inlinings</option></primary></indexterm>
145 <para>inlining info from the simplifier</para>
151 <option>-ddump-cpranal</option>:
152 <indexterm><primary><option>-ddump-cpranal</option></primary></indexterm>
155 <para>CPR analyser output</para>
161 <option>-ddump-stranal</option>:
162 <indexterm><primary><option>-ddump-stranal</option></primary></indexterm>
165 <para>strictness analyser output</para>
171 <option>-ddump-cse</option>:
172 <indexterm><primary><option>-ddump-cse</option></primary></indexterm>
175 <para>CSE pass output</para>
181 <option>-ddump-workwrap</option>:
182 <indexterm><primary><option>-ddump-workwrap</option></primary></indexterm>
185 <para>worker/wrapper split output</para>
191 <option>-ddump-occur-anal</option>:
192 <indexterm><primary><option>-ddump-occur-anal</option></primary></indexterm>
195 <para>`occurrence analysis' output</para>
201 <option>-ddump-prep</option>:
202 <indexterm><primary><option>-ddump-prep</option></primary></indexterm>
205 <para>output of core preparation pass</para>
211 <option>-ddump-stg</option>:
212 <indexterm><primary><option>-ddump-stg</option></primary></indexterm>
215 <para>output of STG-to-STG passes</para>
221 <option>-ddump-flatC</option>:
222 <indexterm><primary><option>-ddump-flatC</option></primary></indexterm>
225 <para><emphasis>flattened</emphasis> Abstract C</para>
231 <option>-ddump-cmm</option>:
232 <indexterm><primary><option>-ddump-cmm</option></primary></indexterm>
235 <para>Print the C-- code out.</para>
241 <option>-ddump-opt-cmm</option>:
242 <indexterm><primary><option>-ddump-opt-cmm</option></primary></indexterm>
245 <para>Dump the results of C-- to C-- optimising passes.</para>
251 <option>-ddump-asm</option>:
252 <indexterm><primary><option>-ddump-asm</option></primary></indexterm>
255 <para>assembly language from the native-code generator</para>
261 <option>-ddump-bcos</option>:
262 <indexterm><primary><option>-ddump-bcos</option></primary></indexterm>
265 <para>byte code compiler output</para>
271 <option>-ddump-foreign</option>:
272 <indexterm><primary><option>-ddump-foreign</option></primary></indexterm>
275 <para>dump foreign export stubs</para>
284 <option>-ddump-simpl-phases</option>:
285 <indexterm><primary><option>-ddump-simpl-phases</option></primary></indexterm>
288 <para>Show the output of each run of the simplifier. Used when even
289 <option>-dverbose-simpl</option> doesn't cut it.</para>
295 <option>-ddump-simpl-iterations</option>:
296 <indexterm><primary><option>-ddump-simpl-iterations</option></primary></indexterm>
299 <para>Show the output of each <emphasis>iteration</emphasis>
300 of the simplifier (each run of the simplifier has a maximum
301 number of iterations, normally 4). This outputs even more information
302 than <option>-ddump-simpl-phases</option>.</para>
308 <option>-ddump-simpl-stats</option>
309 <indexterm><primary><option>-ddump-simpl-stats option</option></primary></indexterm>
312 <para>Dump statistics about how many of each kind of
313 transformation too place. If you add
314 <option>-dppr-debug</option> you get more detailed
321 <option>-ddump-if-trace</option>
322 <indexterm><primary><option>-ddump-if-trace</option></primary></indexterm>
325 <para>Make the interface loader be *real* chatty about what it is
332 <option>-ddump-tc-trace</option>
333 <indexterm><primary><option>-ddump-tc-trace</option></primary></indexterm>
336 <para>Make the type checker be *real* chatty about what it is
343 <option>-ddump-rn-trace</option>
344 <indexterm><primary><option>-ddump-rn-trace</option></primary></indexterm>
347 <para>Make the renamer be *real* chatty about what it is
354 <option>-ddump-rn-stats</option>
355 <indexterm><primary><option>-dshow-rn-stats</option></primary></indexterm>
358 <para>Print out summary of what kind of information the renamer
359 had to bring in.</para>
365 <option>-dverbose-core2core</option>
366 <indexterm><primary><option>-dverbose-core2core</option></primary></indexterm>
369 <option>-dverbose-stg2stg</option>
370 <indexterm><primary><option>-dverbose-stg2stg</option></primary></indexterm>
373 <para>Show the output of the intermediate Core-to-Core and
374 STG-to-STG passes, respectively. (<emphasis>Lots</emphasis>
375 of output!) So: when we're really desperate:</para>
378 % ghc -noC -O -ddump-simpl -dverbose-simpl -dcore-lint Foo.hs
386 <option>-dshow-passes</option>
387 <indexterm><primary><option>-dshow-passes</option></primary></indexterm>
390 <para>Print out each pass name as it happens.</para>
396 <option>-dfaststring-stats</option>
397 <indexterm><primary><option>-dfaststring-stats</option></primary></indexterm>
400 <para>Show statistics for the usage of fast strings by the
407 <option>-dppr-debug</option>
408 <indexterm><primary><option>-dppr-debug</option></primary></indexterm>
411 <para>Debugging output is in one of several
412 “styles.” Take the printing of types, for
413 example. In the “user” style (the default), the
414 compiler's internal ideas about types are presented in
415 Haskell source-level syntax, insofar as possible. In the
416 “debug” style (which is the default for
417 debugging output), the types are printed in with explicit
418 foralls, and variables have their unique-id attached (so you
419 can check for things that look the same but aren't). This
420 flag makes debugging output appear in the more verbose debug
427 <option>-dsuppress-uniques</option>
428 <indexterm><primary><option>-dsuppress-uniques</option></primary></indexterm>
431 <para>Suppress the printing of uniques in debugging output. This may make
432 the printout ambiguous (e.g. unclear where an occurrence of 'x' is bound), but
433 it makes the output of two compiler runs have many fewer gratuitous differences,
434 so you can realistically apply <command>diff</command>. Once <command>diff</command>
435 has shown you where to look, you can try again without <option>-dsuppress-uniques</option></para>
441 <option>-dsuppress-coercions</option>
442 <indexterm><primary><option>-dsuppress-coercions</option></primary></indexterm>
445 <para>Suppress the printing of coercions in Core dumps to make them
452 <option>-dppr-user-length</option>
453 <indexterm><primary><option>-dppr-user-length</option></primary></indexterm>
456 <para>In error messages, expressions are printed to a
457 certain “depth”, with subexpressions beyond the
458 depth replaced by ellipses. This flag sets the
459 depth. Its default value is 5.</para>
465 <option>-dno-debug-output</option>
466 <indexterm><primary><option>-dno-debug-output</option></primary></indexterm>
469 <para>Suppress any unsolicited debugging output. When GHC
470 has been built with the <literal>DEBUG</literal> option it
471 occasionally emits debug output of interest to developers.
472 The extra output can confuse the testing framework and
473 cause bogus test failures, so this flag is provided to
480 <sect2 id="checking-consistency">
481 <title>Checking for consistency</title>
483 <indexterm><primary>consistency checks</primary></indexterm>
484 <indexterm><primary>lint</primary></indexterm>
490 <option>-dcore-lint</option>
491 <indexterm><primary><option>-dcore-lint</option></primary></indexterm>
494 <para>Turn on heavyweight intra-pass sanity-checking within
495 GHC, at Core level. (It checks GHC's sanity, not yours.)</para>
501 <option>-dstg-lint</option>:
502 <indexterm><primary><option>-dstg-lint</option></primary></indexterm>
505 <para>Ditto for STG level. (NOTE: currently doesn't work).</para>
511 <option>-dcmm-lint</option>:
512 <indexterm><primary><option>-dcmm-lint</option></primary></indexterm>
515 <para>Ditto for C-- level.</para>
523 <title>How to read Core syntax (from some <option>-ddump</option>
526 <indexterm><primary>reading Core syntax</primary></indexterm>
527 <indexterm><primary>Core syntax, how to read</primary></indexterm>
529 <para>Let's do this by commenting an example. It's from doing
530 <option>-ddump-ds</option> on this code:
533 skip2 m = m : skip2 (m+2)
536 Before we jump in, a word about names of things. Within GHC,
537 variables, type constructors, etc., are identified by their
538 “Uniques.” These are of the form `letter' plus
539 `number' (both loosely interpreted). The `letter' gives some idea
540 of where the Unique came from; e.g., <literal>_</literal>
541 means “built-in type variable”; <literal>t</literal>
542 means “from the typechecker”; <literal>s</literal>
543 means “from the simplifier”; and so on. The `number'
544 is printed fairly compactly in a `base-62' format, which everyone
545 hates except me (WDP).</para>
547 <para>Remember, everything has a “Unique” and it is
548 usually printed out when debugging, in some form or another. So
549 here we go…</para>
553 Main.skip2{-r1L6-} :: _forall_ a$_4 =>{{Num a$_4}} -> a$_4 -> [a$_4]
555 --# `r1L6' is the Unique for Main.skip2;
556 --# `_4' is the Unique for the type-variable (template) `a'
557 --# `{{Num a$_4}}' is a dictionary argument
561 --# `_NI_' means "no (pragmatic) information" yet; it will later
562 --# evolve into the GHC_PRAGMA info that goes into interface files.
565 /\ _4 -> \ d.Num.t4Gt ->
568 +.t4Hg :: _4 -> _4 -> _4
570 +.t4Hg = (+{-r3JH-} _4) d.Num.t4Gt
572 fromInt.t4GS :: Int{-2i-} -> _4
574 fromInt.t4GS = (fromInt{-r3JX-} _4) d.Num.t4Gt
576 --# The `+' class method (Unique: r3JH) selects the addition code
577 --# from a `Num' dictionary (now an explicit lambda'd argument).
578 --# Because Core is 2nd-order lambda-calculus, type applications
579 --# and lambdas (/\) are explicit. So `+' is first applied to a
580 --# type (`_4'), then to a dictionary, yielding the actual addition
581 --# function that we will use subsequently...
583 --# We play the exact same game with the (non-standard) class method
584 --# `fromInt'. Unsurprisingly, the type `Int' is wired into the
594 } in fromInt.t4GS ds.d4Qz
596 --# `I# 2#' is just the literal Int `2'; it reflects the fact that
597 --# GHC defines `data Int = I# Int#', where Int# is the primitive
598 --# unboxed type. (see relevant info about unboxed types elsewhere...)
600 --# The `!' after `I#' indicates that this is a *saturated*
601 --# application of the `I#' data constructor (i.e., not partially
604 skip2.t3Ja :: _4 -> [_4]
608 let { ds.d4QQ :: [_4]
614 ds.d4QY = +.t4Hg m.r1H4 lit.t4Hb
615 } in skip2.t3Ja ds.d4QY
623 <para>(“It's just a simple functional language” is an
624 unregisterised trademark of Peyton Jones Enterprises, plc.)</para>
629 <title>Unregisterised compilation</title>
630 <indexterm><primary>unregisterised compilation</primary></indexterm>
632 <para>The term "unregisterised" really means "compile via vanilla
633 C", disabling some of the platform-specific tricks that GHC
634 normally uses to make programs go faster. When compiling
635 unregisterised, GHC simply generates a C file which is compiled
638 <para>Unregisterised compilation can be useful when porting GHC to
639 a new machine, since it reduces the prerequisite tools to
640 <command>gcc</command>, <command>as</command>, and
641 <command>ld</command> and nothing more, and furthermore the amount
642 of platform-specific code that needs to be written in order to get
643 unregisterised compilation going is usually fairly small.</para>
645 <para>Unregisterised compilation cannot be selected at
646 compile-time; you have to build GHC with the appropriate options
647 set. Consult the GHC Building Guide for details.</para>
653 ;;; Local Variables: ***
655 ;;; sgml-parent-document: ("users_guide.xml" "book" "chapter" "sect1") ***