1. 12 Nov, 2015 4 commits
  2. 11 Nov, 2015 8 commits
    • Matthew Pickering's avatar
      Remove redundant test. · a038b724
      Matthew Pickering authored
      ExportSyntaxImport will fail if this test would have ever failed.
      a038b724
    • Matthew Pickering's avatar
      Rename bundled pattern synonym tests to reflect new terminology · 63cad5d4
      Matthew Pickering authored
      This also fixes a bug which causes intermittent test failures due to
      interdependent tests.
      63cad5d4
    • Peter Trommler's avatar
      nativeGen.PPC: Fix shift arith. right > 31 bits · fb0d5120
      Peter Trommler authored and Ben Gamari's avatar Ben Gamari committed
      Arithmetic right shifts of more than 31 bits set all bits to
      the sign bit on PowerPC. iThe assembler does not allow shift
      amounts larger than 32 so do an arithemetic right shift of 31
      bit instead.
      
      Fixes #10870
      
      Test Plan: validate (especially on powerpc)
      
      Reviewers: austin, erikd, bgamari
      
      Reviewed By: bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1459
      
      GHC Trac Issues: #10870
      fb0d5120
    • Sylvain HENRY's avatar
      Detect invalid foreign imports in bytecode compiler · badf5d54
      Sylvain HENRY authored and Ben Gamari's avatar Ben Gamari committed
      The bytecode compiler doesn't handle every foreign import calling
      convention. Instead of crashing during the generation of the foreign
      call, we display an error.
      
      Fix lint warnings
      
      Test Plan: prog014 ghci test added
      
      Reviewers: austin, hvr, bgamari
      
      Reviewed By: bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1458
      
      GHC Trac Issues: #10462
      badf5d54
    • niteria's avatar
      Put kind variables before type variables when specializing · 0f495083
      niteria authored and Ben Gamari's avatar Ben Gamari committed
      When you reverse the order of uniques you get the core lint
      error from the testcase. The testcase is copied from
      tests/simplCore/should_compile/T10689a.hs.
      
      The problem is that we would put type and kind variables ordered by
      unique order, which happened to be the right order for this testcase to
      pass under normal conditions.
      
      I think it's enough to sort them with `sortQuantVars`, but I'm not
      really sure if some more sophisticated dependency analysis isn't needed.
      
      Test Plan: added a new testcase
      
      Reviewers: simonpj, goldfire, simonmar, austin, bgamari
      
      Reviewed By: bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1457
      0f495083
    • Sylvain HENRY's avatar
      Systools: read ELF section without calling readelf · 109d7ce8
      Sylvain HENRY authored and Ben Gamari's avatar Ben Gamari committed
      This patch tackles two issues:
      
      1) GHC stores a "link info" string into a ELF section. Initially a
      section with type "note" was used but GHC didn't follow the ELF
      specification which specifies a record-based format for these sections.
      With D1375 we switched to a "progbits" section type for which there
      isn't any format constraint. This is an issue for D1242 which use GCC's
      --gc-sections which collects "unused" sections, such as our section
      containing link info... In this patch, we fall back to a section with
      type "note" but we respect the specified format.
      
      2) Reading back the ELF section was done by parsing the result of a
      call to "readelf". Calling readelf is problematic because the program
      may not be available or it may be renamed on some platforms (see
      D1326). Moreover we have no garanty that its output layout will stay
      the same in future releases of readelf. Finally we would need to fix
      the parsing to support  "note" sections because of 1. Instead, this
      patch proposes to use Data.Binary.Get to directly read the "link info"
      note into its section. ELF has a specification, hence it should work on
      every conforming platform.
      
      This patch "reverts" D1375, hence it supersedes D1432. It makes D1326
      not necessary anymore.
      
      Test Plan:
      - recomp011 should pass (test that relinking is avoided when both "link
      info" match)
      - we should add a test for ELF objects with more than 0xff00 sections
      => added test "recomp015"
      - we should check that GAS generates 32-bit words with .int on every
      supported platform using ELF (or find a place where this is
      documented). harbomaster and I (@hsyl20) only tested on x86-64. On
      platforms where it is not true, it should make recomp011 fail. =>
      tested to work on Linux/amd64, Solaris/i386 and OpenBSD/amd64
      
      Reviewers: olsner, ony, thomie, kgardas, austin, bgamari
      
      Reviewed By: thomie, bgamari
      
      Subscribers: kgardas, rwbarton, thomie
      
      Differential Revision: https://phabricator.haskell.org/D1381
      
      GHC Trac Issues: #10974, #11022
      109d7ce8
    • thomie's avatar
      OPTIONS_GHC compiler flags may contain spaces (#4931) · fbc2537c
      thomie authored and Ben Gamari's avatar Ben Gamari committed
      When a .hsc contains `#define FOO "bar baz"`, hsc2hs emits:
      
          {-# OPTIONS_GHC -optc-DFOO="bar baz" #-}
      
      Make sure GHC can compile this, by tweaking `HeaderInfo.getOptions` a
      bit.
      
      Test Plan: driver/T4931
      
      Reviewers: austin, bgamari
      
      Reviewed By: bgamari
      
      Differential Revision: https://phabricator.haskell.org/D1452
      
      GHC Trac Issues: #4931
      fbc2537c
    • Matthew Pickering's avatar
      Associate pattern synonyms with types in module exports · 96621b1b
      Matthew Pickering authored and Ben Gamari's avatar Ben Gamari committed
      This patch implements #10653.
      
      It adds the ability to bundle pattern synonyms with type constructors in
      export lists so that users can treat pattern synonyms more like data
      constructors.
      
      Updates haddock submodule.
      
      Test Plan: ./validate
      
      Reviewers: goldfire, austin, bgamari
      
      Reviewed By: bgamari
      
      Subscribers: simonpj, gridaphobe, thomie
      
      Differential Revision: https://phabricator.haskell.org/D1258
      
      GHC Trac Issues: #10653
      96621b1b
  3. 09 Nov, 2015 1 commit
  4. 08 Nov, 2015 1 commit
    • Tamar Christina's avatar
      Fix sporadic failing ghci/Linker/Dyn tests · f4056324
      Tamar Christina authored
      Summary:
      Multiple tests use the same source C file.
      GHC was previously writing the resulting .o files
      in the same folder as the source. When these tests
      are run in parallel, sometimes one of the calls would
      use the partial .o file and throw an error.
      
      The .o files are moved into different folders with this
      change.
      
      Test Plan: make test -C testsuite/tests/ghci/linking/dyn
      
      Reviewers: thomie, austin, bgamari
      
      Reviewed By: bgamari
      
      Differential Revision: https://phabricator.haskell.org/D1449
      f4056324
  5. 07 Nov, 2015 4 commits
    • thomie's avatar
      Parser: allow empty multi-line deprecation warnings · 8262c954
      thomie authored and Ben Gamari's avatar Ben Gamari committed
      This should work,
      
          {-# DEPRECATED someFunction [] #-}
      
      Test Plan: parser/should_compile/T3303
      
      Reviewers: bgamari, austin
      
      Reviewed By: austin
      
      Subscribers: mpickering
      
      Differential Revision: https://phabricator.haskell.org/D1433
      
      GHC Trac Issues: #11044
      8262c954
    • johnleo's avatar
      fix #10734 by adding braces to pretty-printing of let inside do · be885857
      johnleo authored and eir@cis.upenn.edu's avatar eir@cis.upenn.edu committed
      Test Plan: validate
      
      Reviewers: bgamari, austin, goldfire
      
      Reviewed By: goldfire
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1448
      
      GHC Trac Issues: #10734
      be885857
    • Tamar Christina's avatar
      Allow the GHCi Linker to resolve related dependencies when loading DLLs · 6e6438e1
      Tamar Christina authored
      Summary:
      GHCi does not correctly tell the Windows Loader how to handle dependencies to DLL's
      that are not on the standard Windows load path:
      
      1. The directory from which the application loaded.
      2. The current directory.
      3. The system directory. Use the GetSystemDirectory function to get the path of this directory.
      4. The 16-bit system directory. There is no function that obtains the path of this directory,
         but it is searched.
      5. The Windows directory. Use the GetWindowsDirectory function to get the path of this directory.
      6. The directories that are listed in the PATH environment variable.
         Note that this does not include the per-application path specified by the
         AppPaths registry key. The App Paths key is not used when computing the DLL search path.
      
      So what this means is given two DLLs `A` and `B` and `B` depending on `A`.
      If we put both DLLs into a new folder bin and then call GHC with:
      
      `ghc -L$(PWD)/bin -lB`
      
      the loading will fail as the Windows loader will try to load the dependency of `B` and fail
      since it cannot find `A`.
      
      *IMPORTANT* this patch drops XP Support.
      The  APIs being used were natively added to Windows 8+ and backported to Windows 7 and Vista
      via a mandatory security patch (in 2011). This means that there is a chance that KB2533623 has
      not been installed on certain machines. For those machines I display a warning and
      temporarily expand the `PATH` to allow it to load.
      
      This patch will make sure that paths provided by the user with `-L` *and* the folder in which a
      DLL is found are added to the search path. It does so using one of two methods depending upon how
      new of a Windows version we are running on:
      
      - If the APIs are available it will use `addDllDirectory` and `removeDllDirectory`.
         The order of which these directories are searched is nondeterministic.
      - If the APIs are not available it means that we're running on a pretty old unpatched machine.
        But if it's being used in an environment with no internet access it may be the case.
        So if the APIs are not available we temporarily extend the `PATH` with the directories.
        A warning is also displayed to the user informing them that the linking may fail,
        and if it does, install the needed patch. The `PATH` variable has limitations.
      
      Test Plan:
      ./validate
      
      Added two new test T10955 and T10955dyn
      
      Reviewers: erikd, bgamari, thomie, hvr, austin
      
      Reviewed By: erikd, thomie
      
      Subscribers: #ghc_windows_task_force
      
      Differential Revision: https://phabricator.haskell.org/D1340
      
      GHC Trac Issues: #10955
      6e6438e1
    • Simon Marlow's avatar
      Make GHCi & TH work when the compiler is built with -prof · ce1f1607
      Simon Marlow authored
      Summary:
      Amazingly, there were zero changes to the byte code generator and very
      few changes to the interpreter - mainly because we've used good
      abstractions that hide the differences between profiling and
      non-profiling.  So that bit was pleasantly straightforward, but there
      were a pile of other wibbles to get the whole test suite through.
      
      Note that a compiler built with -prof is now like one built with
      -dynamic, in that to use TH you have to build the code the same way.
      For dynamic, we automatically enable -dynamic-too when TH is required,
      but we don't have anything equivalent for profiling, so you have to
      explicitly use -prof when building code that uses TH with a profiled
      compiler.  For this reason Cabal won't work with TH.  We don't expect
      to ship a profiled compiler, so I think that's OK.
      
      Test Plan: validate with GhcProfiled=YES in validate.mk
      
      Reviewers: goldfire, bgamari, rwbarton, austin, hvr, erikd, ezyang
      
      Reviewed By: ezyang
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1407
      
      GHC Trac Issues: #4837, #545
      ce1f1607
  6. 04 Nov, 2015 1 commit
  7. 01 Nov, 2015 5 commits
  8. 31 Oct, 2015 1 commit
  9. 30 Oct, 2015 12 commits
    • Sergei Trofimovich's avatar
    • Sergei Trofimovich's avatar
    • niteria's avatar
      Make type-class dictionary let binds deterministic · a5cb27f3
      niteria authored and Ben Gamari's avatar Ben Gamari committed
      When generating dictionary let binds in dsTcEvBinds we may
      end up generating them in arbitrary order according to Unique order.
      
      Consider:
      
      ```
      let $dEq = GHC.Classes.$fEqInt in
      let $$dNum = GHC.Num.$fNumInt in ...
      ```
      
      vs
      
      ```
      let $dNum = GHC.Num.$fNumInt in
      let $dEq = GHC.Classes.$fEqInt in ...
      ```
      
      The way this change fixes it is by using `UniqDFM` - a type of
      deterministic finite maps of things keyed on `Unique`s. This way when
      you pull out evidence variables corresponding to type-class dictionaries
      they are in deterministic order.
      
      Currently it's the order of insertion and the way it's implemented is by
      tagging the values with the time of insertion.
      
      Test Plan:
      I've added a new test case to reproduce the issue.
      ./validate
      
      Reviewers: ezyang, simonmar, austin, simonpj, bgamari
      
      Reviewed By: simonmar, simonpj, bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1396
      
      GHC Trac Issues: #4012
      a5cb27f3
    • Matthew Pickering's avatar
      Add failing test for #11039 · fce758c5
      Matthew Pickering authored and Ben Gamari's avatar Ben Gamari committed
      Reviewers: austin, bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1406
      
      GHC Trac Issues: #11039
      fce758c5
    • Edward Z. Yang's avatar
      Reimplement shadowing on a per database basis. · 39b71e81
      Edward Z. Yang authored
      
      
      Summary:
      This commit reimplements shadowing on package databases by doing
      the shadowing calculation on a per-database basis: specifically,
      if a later package database shadows a package from the earlier
      databases, we first remove that package (and its transitive
      dependencies) before merging the databases together.
      
      This should also fix bootstrapping GHC HEAD with HEAD.
      Signed-off-by: default avatarEdward Z. Yang <ezyang@cs.stanford.edu>
      
      Test Plan: validate
      
      Reviewers: ggreif, bgamari, austin
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1385
      39b71e81
    • Ben Gamari's avatar
      Generate Typeable info at definition sites · 91c6b1f5
      Ben Gamari authored and Ben Gamari's avatar Ben Gamari committed
      This is the second attempt at merging D757.
      
      This patch implements the idea floated in Trac #9858, namely that we
      should generate type-representation information at the data type
      declaration site, rather than when solving a Typeable constraint.
      
      However, this turned out quite a bit harder than I expected. I still
      think it's the right thing to do, and it's done now, but it was quite
      a struggle.
      
      See particularly
      
       * Note [Grand plan for Typeable] in TcTypeable (which is a new module)
       * Note [The overall promotion story] in DataCon (clarifies existing
      stuff)
      
      The most painful bit was that to generate Typeable instances (ie
      TyConRepName bindings) for every TyCon is tricky for types in ghc-prim
      etc:
      
       * We need to have enough data types around to *define* a TyCon
       * Many of these types are wired-in
      
      Also, to minimise the code generated for each data type, I wanted to
      generate pure data, not CAFs with unpackCString# stuff floating about.
      
      Performance
      ~~~~~~~~~~~
      Three perf/compiler tests start to allocate quite a bit more. This isn't
      surprising, because they all allocate zillions of data types, with
      practically no other code, esp. T1969
      
       * T1969:    GHC allocates 19% more
       * T4801:    GHC allocates 13% more
       * T5321FD:  GHC allocates 13% more
       * T9675:    GHC allocates 11% more
       * T783:     GHC allocates 11% more
       * T5642:    GHC allocates 10% more
      
      I'm treating this as acceptable. The payoff comes in Typeable-heavy
      code.
      
      Remaining to do
      ~~~~~~~~~~~~~~~
      
       * I think that "TyCon" and "Module" are over-generic names to use for
         the runtime type representations used in GHC.Typeable. Better might
      be
         "TrTyCon" and "TrModule". But I have not yet done this
      
       * Add more info the the "TyCon" e.g. source location where it was
         defined
      
       * Use the new "Module" type to help with Trac Trac #10068
      
       * It would be possible to generate TyConRepName (ie Typeable
         instances) selectively rather than all the time. We'd need to persist
         the information in interface files. Lacking a motivating reason I
      have
         not done this, but it would not be difficult.
      
      Refactoring
      ~~~~~~~~~~~
      As is so often the case, I ended up refactoring more than I intended.
      In particular
      
       * In TyCon, a type *family* (whether type or data) is repesented by a
         FamilyTyCon
           * a algebraic data type (including data/newtype instances) is
             represented by AlgTyCon This wasn't true before; a data family
             was represented as an AlgTyCon. There are some corresponding
             changes in IfaceSyn.
      
           * Also get rid of the (unhelpfully named) tyConParent.
      
       * In TyCon define 'Promoted', isomorphic to Maybe, used when things are
         optionally promoted; and use it elsewhere in GHC.
      
       * Cleanup handling of knownKeyNames
      
       * Each TyCon, including promoted TyCons, contains its TyConRepName, if
         it has one. This is, in effect, the name of its Typeable instance.
      
      Updates haddock submodule
      
      Test Plan: Let Harbormaster validate
      
      Reviewers: austin, hvr, goldfire
      
      Subscribers: goldfire, thomie
      
      Differential Revision: https://phabricator.haskell.org/D1404
      
      GHC Trac Issues: #9858
      91c6b1f5
    • thomie's avatar
      Testsuite: suggest quoting $(TEST_HC) · 59e728bc
      thomie authored and Ben Gamari's avatar Ben Gamari committed
      Test Plan: it works
      
      Reviewers: bgamari, rwbarton, austin
      
      Reviewed By: austin
      
      Subscribers: rwbarton
      
      Differential Revision: https://phabricator.haskell.org/D1377
      59e728bc
    • Ben Gamari's avatar
      CmmParse: Expose popcnt operations · 42e85284
      Ben Gamari authored and Ben Gamari's avatar Ben Gamari committed
      Make various population count operations available via C-- syntax
      under the names %popcnt{8,16,32,64}. Fixes #11037.
      
      Reviewers: simonmar, austin, ekmett
      
      Reviewed By: austin, ekmett
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1402
      
      GHC Trac Issues: #11037
      42e85284
    • Adam Gundry's avatar
      Disambiguate record selectors by type signature · 0a163741
      Adam Gundry authored
      This makes DuplicateRecordFields more liberal in when it will
      accept ambiguous record selectors, making use of type information in a
      similar way to updates. See Note [Disambiguating record fields] for more
      details. I've also refactored how record updates are disambiguated.
      
      Test Plan: New and amended tests in overloadedrecflds
      
      Reviewers: simonpj, goldfire, bgamari, austin
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1391
      0a163741
    • thomie's avatar
    • Simon Peyton Jones's avatar
      Record usage information using GlobalRdrElt · 3e94842d
      Simon Peyton Jones authored
      This patch implements an improvment that I've wanted to do for ages, but
      never gotten around to.
      
      Unused imports are computed based on how imported entities occur (qualified,
      unqualified).   This info was accumulated in tcg_used_rdrnames :: Set RdrName.
      But that was a huge pain, and it got worse when we introduced duplicate
      record fields.
      
      The Right Thing is to record tcg_used_gres :: [GlobalRdrElt], which records
      the GRE *after* filtering with pickGREs.  See Note [GRE filtering] in RdrName.
      This is much, much bette.  This patch deletes quite a bit of code, and is
      conceptually much easier to follow.
      
      Hooray.  There should be no change in functionality.
      3e94842d
    • Simon Peyton Jones's avatar
      Fix unused-import stuff in a better way · 9376249b
      Simon Peyton Jones authored
      The fix for Trac #10890 in commit 1818b48e, namely
         Fix incorrect import warnings when methods with identical names are imported
      was wrong, as demonstrated by the new test T10890_2.  It suppressed
      far too many warnings!
      
      This patch fixes the original problem in a different way, by making
      RdrName.greUsedRdrName a bit cleverer.
      
      But this too is not really the Right Thing.  I think the Right Thing is
      to store the /GRE/ in the tcg_used_rdrnames, not the /RdrName/.  That
      would be a lot simpler and more direct.
      
      But one step at a time.
      9376249b
  10. 29 Oct, 2015 3 commits
    • Sergei Trofimovich's avatar
      x86 codegen: don't generate location comments · e272ab99
      Sergei Trofimovich authored
      
      
      The following source snippet 'module A where x */* y = 42'
      when being compiled with '-g' option emits syntactically
      invalid comment for GNU as:
      
          .text
              .align 8
              .loc 1 3 1 /* */* */
      
      Fixed by not emitting comments at all. We already suppress
      all asm comments in 'X86/Ppr.hs'.
      Signed-off-by: default avatarSergei Trofimovich <siarheit@google.com>
      
      Test Plan: added test and check it works
      
      Reviewers: scpmw, simonmar, austin, bgamari
      
      Reviewed By: simonmar
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D1386
      
      GHC Trac Issues: #10667
      e272ab99
    • Ben Gamari's avatar
      Revert "Generate Typeable info at definition sites" · bbaf76f9
      Ben Gamari authored
      This reverts commit bef2f03e.
      
      This merge was botched
      
      Also reverts haddock submodule.
      bbaf76f9
    • Ben Gamari's avatar
      Generate Typeable info at definition sites · bef2f03e
      Ben Gamari authored
      This patch implements the idea floated in Trac #9858, namely that we
      should generate type-representation information at the data type
      declaration site, rather than when solving a Typeable constraint.
      
      However, this turned out quite a bit harder than I expected. I still
      think it's the right thing to do, and it's done now, but it was quite
      a struggle.
      
      See particularly
      
       * Note [Grand plan for Typeable] in TcTypeable (which is a new module)
       * Note [The overall promotion story] in DataCon (clarifies existing stuff)
      
      The most painful bit was that to generate Typeable instances (ie
      TyConRepName bindings) for every TyCon is tricky for types in ghc-prim
      etc:
      
       * We need to have enough data types around to *define* a TyCon
       * Many of these types are wired-in
      
      Also, to minimise the code generated for each data type, I wanted to
      generate pure data, not CAFs with unpackCString# stuff floating about.
      
      Performance
      ~~~~~~~~~~~
      Three perf/compiler tests start to allocate quite a bit more. This isn't
      surprising, because they all allocate zillions of data types, with
      practically no other code, esp. T1969
      
       * T3294:   GHC allocates 110% more (filed #11030 to track this)
       * T1969:   GHC allocates 30% more
       * T4801:   GHC allocates 14% more
       * T5321FD: GHC allocates 13% more
       * T783:    GHC allocates 12% more
       * T9675:   GHC allocates 12% more
       * T5642:   GHC allocates 10% more
       * T9961:   GHC allocates 6% more
      
       * T9203:   Program allocates 54% less
      
      I'm treating this as acceptable. The payoff comes in Typeable-heavy
      code.
      
      Remaining to do
      ~~~~~~~~~~~~~~~
      
       * I think that "TyCon" and "Module" are over-generic names to use for
         the runtime type representations used in GHC.Typeable. Better might be
         "TrTyCon" and "TrModule". But I have not yet done this
      
       * Add more info the the "TyCon" e.g. source location where it was
         defined
      
       * Use the new "Module" type to help with Trac Trac #10068
      
       * It would be possible to generate TyConRepName (ie Typeable
         instances) selectively rather than all the time. We'd need to persist
         the information in interface files. Lacking a motivating reason I have
         not done this, but it would not be difficult.
      
      Refactoring
      ~~~~~~~~~~~
      As is so often the case, I ended up refactoring more than I intended.
      In particular
      
       * In TyCon, a type *family* (whether type or data) is repesented by a
         FamilyTyCon
           * a algebraic data type (including data/newtype instances) is
             represented by AlgTyCon This wasn't true before; a data family
             was represented as an AlgTyCon. There are some corresponding
             changes in IfaceSyn.
      
           * Also get rid of the (unhelpfully named) tyConParent.
      
       * In TyCon define 'Promoted', isomorphic to Maybe, used when things are
         optionally promoted; and use it elsewhere in GHC.
      
       * Cleanup handling of knownKeyNames
      
       * Each TyCon, including promoted TyCons, contains its TyConRepName, if
         it has one. This is, in effect, the name of its Typeable instance.
      
      Requires update of the haddock submodule.
      
      Differential Revision: https://phabricator.haskell.org/D757
      bef2f03e