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-tc</option>:
64 <indexterm><primary><option>-ddump-tc</option></primary></indexterm>
67 <para>typechecker output</para>
73 <option>-ddump-splices</option>:
74 <indexterm><primary><option>-ddump-splices</option></primary></indexterm>
77 <para>Dump Template Haskell expressions that we splice in,
78 and what Haskell code the expression evaluates to.</para>
84 <option>-ddump-types</option>:
85 <indexterm><primary><option>-ddump-types</option></primary></indexterm>
88 <para>Dump a type signature for each value defined at
89 the top level of the module. The list is sorted
90 alphabetically. Using <option>-dppr-debug</option>
91 dumps a type signature for all the imported and
92 system-defined things as well; useful for debugging the
99 <option>-ddump-deriv</option>:
100 <indexterm><primary><option>-ddump-deriv</option></primary></indexterm>
103 <para>derived instances</para>
109 <option>-ddump-ds</option>:
110 <indexterm><primary><option>-ddump-ds</option></primary></indexterm>
113 <para>desugarer output</para>
119 <option>-ddump-spec</option>:
120 <indexterm><primary><option>-ddump-spec</option></primary></indexterm>
123 <para>output of specialisation pass</para>
129 <option>-ddump-rules</option>:
130 <indexterm><primary><option>-ddump-rules</option></primary></indexterm>
133 <para>dumps all rewrite rules (including those generated
134 by the specialisation pass)</para>
140 <option>-ddump-simpl</option>:
141 <indexterm><primary><option>-ddump-simpl</option></primary></indexterm>
144 <para>simplifier output (Core-to-Core passes)</para>
150 <option>-ddump-inlinings</option>:
151 <indexterm><primary><option>-ddump-inlinings</option></primary></indexterm>
154 <para>inlining info from the simplifier</para>
160 <option>-ddump-cpranal</option>:
161 <indexterm><primary><option>-ddump-cpranal</option></primary></indexterm>
164 <para>CPR analyser output</para>
170 <option>-ddump-stranal</option>:
171 <indexterm><primary><option>-ddump-stranal</option></primary></indexterm>
174 <para>strictness analyser output</para>
180 <option>-ddump-cse</option>:
181 <indexterm><primary><option>-ddump-cse</option></primary></indexterm>
184 <para>CSE pass output</para>
190 <option>-ddump-workwrap</option>:
191 <indexterm><primary><option>-ddump-workwrap</option></primary></indexterm>
194 <para>worker/wrapper split output</para>
200 <option>-ddump-occur-anal</option>:
201 <indexterm><primary><option>-ddump-occur-anal</option></primary></indexterm>
204 <para>`occurrence analysis' output</para>
210 <option>-ddump-prep</option>:
211 <indexterm><primary><option>-ddump-prep</option></primary></indexterm>
214 <para>output of core preparation pass</para>
220 <option>-ddump-stg</option>:
221 <indexterm><primary><option>-ddump-stg</option></primary></indexterm>
224 <para>output of STG-to-STG passes</para>
230 <option>-ddump-flatC</option>:
231 <indexterm><primary><option>-ddump-flatC</option></primary></indexterm>
234 <para><emphasis>flattened</emphasis> Abstract C</para>
240 <option>-ddump-cmm</option>:
241 <indexterm><primary><option>-ddump-cmm</option></primary></indexterm>
244 <para>Print the C-- code out.</para>
250 <option>-ddump-opt-cmm</option>:
251 <indexterm><primary><option>-ddump-opt-cmm</option></primary></indexterm>
254 <para>Dump the results of C-- to C-- optimising passes.</para>
260 <option>-ddump-asm</option>:
261 <indexterm><primary><option>-ddump-asm</option></primary></indexterm>
264 <para>assembly language from the native-code generator</para>
270 <option>-ddump-bcos</option>:
271 <indexterm><primary><option>-ddump-bcos</option></primary></indexterm>
274 <para>byte code compiler output</para>
280 <option>-ddump-foreign</option>:
281 <indexterm><primary><option>-ddump-foreign</option></primary></indexterm>
284 <para>dump foreign export stubs</para>
293 <option>-ddump-simpl-iterations</option>:
294 <indexterm><primary><option>-ddump-simpl-iterations</option></primary></indexterm>
297 <para>Show the output of each <emphasis>iteration</emphasis>
298 of the simplifier (each run of the simplifier has a maximum
299 number of iterations, normally 4). Used when even
300 <option>-dverbose-simpl</option> doesn't cut it.</para>
306 <option>-ddump-simpl-stats</option>
307 <indexterm><primary><option>-ddump-simpl-stats option</option></primary></indexterm>
310 <para>Dump statistics about how many of each kind of
311 transformation too place. If you add
312 <option>-dppr-debug</option> you get more detailed
319 <option>-ddump-if-trace</option>
320 <indexterm><primary><option>-ddump-if-trace</option></primary></indexterm>
323 <para>Make the interface loader be *real* chatty about what it is
330 <option>-ddump-tc-trace</option>
331 <indexterm><primary><option>-ddump-tc-trace</option></primary></indexterm>
334 <para>Make the type checker be *real* chatty about what it is
341 <option>-ddump-rn-trace</option>
342 <indexterm><primary><option>-ddump-rn-trace</option></primary></indexterm>
345 <para>Make the renamer be *real* chatty about what it is
352 <option>-ddump-rn-stats</option>
353 <indexterm><primary><option>-dshow-rn-stats</option></primary></indexterm>
356 <para>Print out summary of what kind of information the renamer
357 had to bring in.</para>
363 <option>-dverbose-core2core</option>
364 <indexterm><primary><option>-dverbose-core2core</option></primary></indexterm>
367 <option>-dverbose-stg2stg</option>
368 <indexterm><primary><option>-dverbose-stg2stg</option></primary></indexterm>
371 <para>Show the output of the intermediate Core-to-Core and
372 STG-to-STG passes, respectively. (<emphasis>Lots</emphasis>
373 of output!) So: when we're really desperate:</para>
376 % ghc -noC -O -ddump-simpl -dverbose-simpl -dcore-lint Foo.hs
384 <option>-dshow-passes</option>
385 <indexterm><primary><option>-dshow-passes</option></primary></indexterm>
388 <para>Print out each pass name as it happens.</para>
394 <option>-dfaststring-stats</option>
395 <indexterm><primary><option>-dfaststring-stats</option></primary></indexterm>
398 <para>Show statistics for the usage of fast strings by the
405 <option>-dppr-debug</option>
406 <indexterm><primary><option>-dppr-debug</option></primary></indexterm>
409 <para>Debugging output is in one of several
410 “styles.” Take the printing of types, for
411 example. In the “user” style (the default), the
412 compiler's internal ideas about types are presented in
413 Haskell source-level syntax, insofar as possible. In the
414 “debug” style (which is the default for
415 debugging output), the types are printed in with explicit
416 foralls, and variables have their unique-id attached (so you
417 can check for things that look the same but aren't). This
418 flag makes debugging output appear in the more verbose debug
425 <option>-dppr-user-length</option>
426 <indexterm><primary><option>-dppr-user-length</option></primary></indexterm>
429 <para>In error messages, expressions are printed to a
430 certain “depth”, with subexpressions beyond the
431 depth replaced by ellipses. This flag sets the
438 <option>-dshow-unused-imports</option>
439 <indexterm><primary><option>-dshow-unused-imports</option></primary></indexterm>
442 <para>Have the renamer report what imports does not
449 <sect2 id="checking-consistency">
450 <title>Checking for consistency</title>
452 <indexterm><primary>consistency checks</primary></indexterm>
453 <indexterm><primary>lint</primary></indexterm>
459 <option>-dcore-lint</option>
460 <indexterm><primary><option>-dcore-lint</option></primary></indexterm>
463 <para>Turn on heavyweight intra-pass sanity-checking within
464 GHC, at Core level. (It checks GHC's sanity, not yours.)</para>
470 <option>-dstg-lint</option>:
471 <indexterm><primary><option>-dstg-lint</option></primary></indexterm>
474 <para>Ditto for STG level. (NOTE: currently doesn't work).</para>
480 <option>-dcmm-lint</option>:
481 <indexterm><primary><option>-dcmm-lint</option></primary></indexterm>
484 <para>Ditto for C-- level.</para>
492 <title>How to read Core syntax (from some <option>-ddump</option>
495 <indexterm><primary>reading Core syntax</primary></indexterm>
496 <indexterm><primary>Core syntax, how to read</primary></indexterm>
498 <para>Let's do this by commenting an example. It's from doing
499 <option>-ddump-ds</option> on this code:
502 skip2 m = m : skip2 (m+2)
505 Before we jump in, a word about names of things. Within GHC,
506 variables, type constructors, etc., are identified by their
507 “Uniques.” These are of the form `letter' plus
508 `number' (both loosely interpreted). The `letter' gives some idea
509 of where the Unique came from; e.g., <literal>_</literal>
510 means “built-in type variable”; <literal>t</literal>
511 means “from the typechecker”; <literal>s</literal>
512 means “from the simplifier”; and so on. The `number'
513 is printed fairly compactly in a `base-62' format, which everyone
514 hates except me (WDP).</para>
516 <para>Remember, everything has a “Unique” and it is
517 usually printed out when debugging, in some form or another. So
518 here we go…</para>
522 Main.skip2{-r1L6-} :: _forall_ a$_4 =>{{Num a$_4}} -> a$_4 -> [a$_4]
524 --# `r1L6' is the Unique for Main.skip2;
525 --# `_4' is the Unique for the type-variable (template) `a'
526 --# `{{Num a$_4}}' is a dictionary argument
530 --# `_NI_' means "no (pragmatic) information" yet; it will later
531 --# evolve into the GHC_PRAGMA info that goes into interface files.
534 /\ _4 -> \ d.Num.t4Gt ->
537 +.t4Hg :: _4 -> _4 -> _4
539 +.t4Hg = (+{-r3JH-} _4) d.Num.t4Gt
541 fromInt.t4GS :: Int{-2i-} -> _4
543 fromInt.t4GS = (fromInt{-r3JX-} _4) d.Num.t4Gt
545 --# The `+' class method (Unique: r3JH) selects the addition code
546 --# from a `Num' dictionary (now an explicit lambda'd argument).
547 --# Because Core is 2nd-order lambda-calculus, type applications
548 --# and lambdas (/\) are explicit. So `+' is first applied to a
549 --# type (`_4'), then to a dictionary, yielding the actual addition
550 --# function that we will use subsequently...
552 --# We play the exact same game with the (non-standard) class method
553 --# `fromInt'. Unsurprisingly, the type `Int' is wired into the
563 } in fromInt.t4GS ds.d4Qz
565 --# `I# 2#' is just the literal Int `2'; it reflects the fact that
566 --# GHC defines `data Int = I# Int#', where Int# is the primitive
567 --# unboxed type. (see relevant info about unboxed types elsewhere...)
569 --# The `!' after `I#' indicates that this is a *saturated*
570 --# application of the `I#' data constructor (i.e., not partially
573 skip2.t3Ja :: _4 -> [_4]
577 let { ds.d4QQ :: [_4]
583 ds.d4QY = +.t4Hg m.r1H4 lit.t4Hb
584 } in skip2.t3Ja ds.d4QY
592 <para>(“It's just a simple functional language” is an
593 unregisterised trademark of Peyton Jones Enterprises, plc.)</para>
598 <title>Unregisterised compilation</title>
599 <indexterm><primary>unregisterised compilation</primary></indexterm>
601 <para>The term "unregisterised" really means "compile via vanilla
602 C", disabling some of the platform-specific tricks that GHC
603 normally uses to make programs go faster. When compiling
604 unregisterised, GHC simply generates a C file which is compiled
607 <para>Unregisterised compilation can be useful when porting GHC to
608 a new machine, since it reduces the prerequisite tools to
609 <command>gcc</command>, <command>as</command>, and
610 <command>ld</command> and nothing more, and furthermore the amount
611 of platform-specific code that needs to be written in order to get
612 unregisterised compilation going is usually fairly small.</para>
617 <option>-unreg</option>:
618 <indexterm><primary><option>-unreg</option></primary></indexterm>
621 <para>Compile via vanilla ANSI C only, turning off
622 platform-specific optimisations. NOTE: in order to use
623 <option>-unreg</option>, you need to have a set of libraries
624 (including the RTS) built for unregisterised compilation.
625 This amounts to building GHC with way "u" enabled.</para>
634 ;;; Local Variables: ***
636 ;;; sgml-parent-document: ("users_guide.xml" "book" "chapter" "sect1") ***