1. 05 Aug, 2016 6 commits
  2. 04 Aug, 2016 3 commits
  3. 03 Aug, 2016 11 commits
  4. 02 Aug, 2016 1 commit
  5. 01 Aug, 2016 3 commits
  6. 30 Jul, 2016 1 commit
    • alexbiehl's avatar
      Add atomic operations to package.conf.in · c06e3f46
      alexbiehl authored
      This patch resulted from the discussion in D2431 and should be merged
      first.
      
      @erikd and @trommler reported errors like
      
      ```
      /home/erikd/Git/ghc-upstream/rts/dist/build/libHSrts_thr.a(PrimOps.thr_o
      ): In function `c14_info':
      (.text+0x2b8): undefined reference to `hs_cmpxchg32'
      /home/erikd/Git/ghc-upstream/rts/dist/build/libHSrts_thr.a(PrimOps.thr_o
      ): In function `c5e_info':
      (.text+0xac4): undefined reference to `hs_cmpxchg32'
      /home/erikd/Git/ghc-upstream/rts/dist/build/libHSrts_thr.a(PrimOps.thr_o
      ): In function `c8b_info':
      (.text+0x1198): undefined reference to `hs_cmpxchg32'
      /home/erikd/Git/ghc-upstream/rts/dist/build/libHSrts_thr.a(PrimOps.thr_o
      ): In function `c8b_info':
      (.text+0x122c): undefined reference to `hs_cmpxchg32'
      /home/erikd/Git/ghc-upstream/rts/dist/build/libHSrts_thr.a(PrimOps.thr_o
      ): In function `c8b_info':
      (.text+0x12ec): undefined reference to `hs_cmpxchg32'
      ```
      
      on PowerPC. @simonmar suggests to add the specific exports to
      `rts/package.conf.in`. This patch does exactly that, including all
      other atomic ops as they probably (maybe someone can verify?) suffer
      from the same problem on PPC.
      
      Test Plan: Please make sure to build on PPC.
      
      Reviewers: erikd, austin, bgamari, simonmar, trommler
      
      Reviewed By: erikd, trommler
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D2435
      c06e3f46
  7. 28 Jul, 2016 1 commit
  8. 27 Jul, 2016 2 commits
  9. 26 Jul, 2016 4 commits
  10. 25 Jul, 2016 1 commit
  11. 24 Jul, 2016 1 commit
  12. 22 Jul, 2016 6 commits
    • Erik de Castro Lopo's avatar
      Fix the non-Linux build · d068220f
      Erik de Castro Lopo authored
      Summary:
      The recent Compact Regions commit (cf989ffe) builds fine on Linux
      but doesn't build on OS X r Windows.
      
      * rts/sm/CNF.c: Drop un-needed #includes.
      * Fix parenthesis usage with CPP ASSERT macro.
      * Fix format string in debugBelch messages.
      * Use stg_max() instead hand rolled inline max() function.
      
      Test Plan: Build on Linux, OS X and Windows
      
      Reviewers: gcampax, simonmar, austin, bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D2421
      d068220f
    • Simon Marlow's avatar
      UNPACK the size field of SizedSeq · 1fe5c891
      Simon Marlow authored
      1fe5c891
    • Simon Marlow's avatar
      -fprof-auto-top · 7f0f1d7f
      Simon Marlow authored
      7f0f1d7f
    • Simon Marlow's avatar
      Squash space leaks in the result of byteCodeGen · 648fd73a
      Simon Marlow authored
      When loading a large number of modules into GHCi, we collect
      CompiledByteCode for every module and then link it all at the end.
      Space leaks in the CompiledByteCode linger until we traverse it all for
      linking, and possibly longer, if there are bits we don't look at.
      
      This is the nuke-it-from-orbit approach: we deepseq the whole thing
      after code generation. It's the only way to be sure.
      
      Test Plan:
      Heap profile of GHCi while loading nofib/real/anna into GHCi, this patch
      reduces the peak heap usage from ~100M to ~50M.
      
      Reviewers: hvr, austin, bgamari, erikd
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D2419
      648fd73a
    • Simon Marlow's avatar
      Add deepseq dependency and a few NFData instances · c4f3d91b
      Simon Marlow authored
      I needed to rnf a data structure (CompiledByteCode) but we don't have
      any good deepseq infrastructure in the compiler yet.  There are bits and
      pieces, but nothing consistent, so this is a start.
      
      We already had a dependency on deepseq indirectly via other packages
      (e.g. containers).
      
      Includes an update to the haddock submodule, to remove orphan NFData
      instances in there.
      
      Test Plan: validate
      
      Reviewers: austin, bgamari, erikd, hvr
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D2418
      c4f3d91b
    • Ömer Sinan Ağacan's avatar
      bfef2eb1