Debugging the compiler

Debugging the compiler debugging options (for GHC) HACKER TERRITORY. HACKER TERRITORY. (You were warned.) Dumping out compiler intermediate structures dumping GHC intermediates intermediate passes, output pass options Make a debugging dump after pass <pass> (may be common enough to need a short form…). You can get all of these at once (lots of output) by using , or most of them with . You can prevent them from clogging up your standard output by passing . Some of the most useful ones are: : parser output : renamer output : typechecker output : Dump Template Haskell expressions that we splice in, and what Haskell code the expression evaluates to. : Dump a type signature for each value defined at the top level of the module. The list is sorted alphabetically. Using dumps a type signature for all the imported and system-defined things as well; useful for debugging the compiler. : derived instances : desugarer output : output of specialisation pass : dumps all rewrite rules specified in this module; see . : dumps the names of all rules that fired in this module : dumps detailed information about all rules that fired in this module : dumps the output of the vectoriser. : simplifier output (Core-to-Core passes) : inlining info from the simplifier : CPR analyser output : strictness analyser output : CSE pass output : worker/wrapper split output : `occurrence analysis' output : output of core preparation pass : output of STG-to-STG passes : flattened Abstract C : Print the C-- code out. : Dump the results of C-- to C-- optimising passes. : assembly language from the native code generator : LLVM code from the LLVM code generator : byte code compiler output : dump foreign export stubs : Show the output of each run of the simplifier. Used when even doesn't cut it. : Show the output of each iteration of the simplifier (each run of the simplifier has a maximum number of iterations, normally 4). This outputs even more information than . Dump statistics about how many of each kind of transformation too place. If you add you get more detailed information. Make the interface loader be *real* chatty about what it is up to. Make the type checker be *real* chatty about what it is up to. Make the vectoriser be *real* chatty about what it is up to. Make the renamer be *real* chatty about what it is up to. Print out summary of what kind of information the renamer had to bring in. Show the output of the intermediate Core-to-Core and STG-to-STG passes, respectively. (Lots of output!) So: when we're really desperate: % ghc -noC -O -ddump-simpl -dverbose-core2core -dcore-lint Foo.hs Print out each pass name as it happens. Print a one-line summary of the size of the Core program at the end of the optimisation pipeline. Show statistics for the usage of fast strings by the compiler. Debugging output is in one of several “styles.” Take the printing of types, for example. In the “user” style (the default), the compiler's internal ideas about types are presented in Haskell source-level syntax, insofar as possible. In the “debug” style (which is the default for debugging output), the types are printed in with explicit foralls, and variables have their unique-id attached (so you can check for things that look the same but aren't). This flag makes debugging output appear in the more verbose debug style. Formatting dumps formatting dumps In error messages, expressions are printed to a certain “depth”, with subexpressions beyond the depth replaced by ellipses. This flag sets the depth. Its default value is 5. Set the width of debugging output. Use this if your code is wrapping too much. For example: . Print single alternative case expressions as though they were strict let expressions. This is helpful when your code does a lot of unboxing. Suppress any unsolicited debugging output. When GHC has been built with the DEBUG option it occasionally emits debug output of interest to developers. The extra output can confuse the testing framework and cause bogus test failures, so this flag is provided to turn it off. Suppressing unwanted information suppression Core dumps contain a large amount of information. Depending on what you are doing, not all of it will be useful. Use these flags to suppress the parts that you are not interested in. Suppress everything that can be suppressed, except for unique ids as this often makes the printout ambiguous. If you just want to see the overall structure of the code, then start here. Suppress the printing of uniques. This may make the printout ambiguous (e.g. unclear where an occurrence of 'x' is bound), but it makes the output of two compiler runs have many fewer gratuitous differences, so you can realistically apply diff. Once diff has shown you where to look, you can try again without Suppress extended information about identifiers where they are bound. This includes strictness information and inliner templates. Using this flag can cut the size of the core dump in half, due to the lack of inliner templates Suppress the printing of module qualification prefixes. This is the Data.List in Data.List.length. Suppress the printing of type signatures. Suppress the printing of type applications. Suppress the printing of type coercions. Checking for consistency consistency checks lint Turn on heavyweight intra-pass sanity-checking within GHC, at Core level. (It checks GHC's sanity, not yours.) : Ditto for STG level. (NOTE: currently doesn't work). : Ditto for C-- level. How to read Core syntax (from some <option>-ddump</option> flags) reading Core syntax Core syntax, how to read Let's do this by commenting an example. It's from doing on this code: skip2 m = m : skip2 (m+2) Before we jump in, a word about names of things. Within GHC, variables, type constructors, etc., are identified by their “Uniques.” These are of the form `letter' plus `number' (both loosely interpreted). The `letter' gives some idea of where the Unique came from; e.g., _ means “built-in type variable”; t means “from the typechecker”; s means “from the simplifier”; and so on. The `number' is printed fairly compactly in a `base-62' format, which everyone hates except me (WDP). Remember, everything has a “Unique” and it is usually printed out when debugging, in some form or another. So here we go… Desugared: Main.skip2{-r1L6-} :: _forall_ a$_4 =>{{Num a$_4}} -> a$_4 -> [a$_4] --# `r1L6' is the Unique for Main.skip2; --# `_4' is the Unique for the type-variable (template) `a' --# `{{Num a$_4}}' is a dictionary argument _NI_ --# `_NI_' means "no (pragmatic) information" yet; it will later --# evolve into the GHC_PRAGMA info that goes into interface files. Main.skip2{-r1L6-} = /\ _4 -> \ d.Num.t4Gt -> let { {- CoRec -} +.t4Hg :: _4 -> _4 -> _4 _NI_ +.t4Hg = (+{-r3JH-} _4) d.Num.t4Gt fromInt.t4GS :: Int{-2i-} -> _4 _NI_ fromInt.t4GS = (fromInt{-r3JX-} _4) d.Num.t4Gt --# The `+' class method (Unique: r3JH) selects the addition code --# from a `Num' dictionary (now an explicit lambda'd argument). --# Because Core is 2nd-order lambda-calculus, type applications --# and lambdas (/\) are explicit. So `+' is first applied to a --# type (`_4'), then to a dictionary, yielding the actual addition --# function that we will use subsequently... --# We play the exact same game with the (non-standard) class method --# `fromInt'. Unsurprisingly, the type `Int' is wired into the --# compiler. lit.t4Hb :: _4 _NI_ lit.t4Hb = let { ds.d4Qz :: Int{-2i-} _NI_ ds.d4Qz = I#! 2# } in fromInt.t4GS ds.d4Qz --# `I# 2#' is just the literal Int `2'; it reflects the fact that --# GHC defines `data Int = I# Int#', where Int# is the primitive --# unboxed type. (see relevant info about unboxed types elsewhere...) --# The `!' after `I#' indicates that this is a *saturated* --# application of the `I#' data constructor (i.e., not partially --# applied). skip2.t3Ja :: _4 -> [_4] _NI_ skip2.t3Ja = \ m.r1H4 -> let { ds.d4QQ :: [_4] _NI_ ds.d4QQ = let { ds.d4QY :: _4 _NI_ ds.d4QY = +.t4Hg m.r1H4 lit.t4Hb } in skip2.t3Ja ds.d4QY } in :! _4 m.r1H4 ds.d4QQ {- end CoRec -} } in skip2.t3Ja (“It's just a simple functional language” is an unregisterised trademark of Peyton Jones Enterprises, plc.)