Ticket #5832 (closed bug: fixed)

Opened 16 months ago

Last modified 16 months ago

PolyKinds03 and PolyKinds13 fail an ASSERT

Reported by: igloo Owned by: simonpj
Priority: normal Milestone: 7.6.1
Component: Compiler Version: 7.4.1-rc2
Keywords: PolyKinds Cc:
Operating System: Unknown/Multiple Architecture: Unknown/Multiple
Type of failure: None/Unknown Difficulty: Unknown
Test Case: PolyKinds03 PolyKinds13 Blocked By:
Blocking: Related Tickets:

Description

With a slow validate build on OSX, PolyKinds03 fails:

cd . && '/Users/ian/ghc/7.4-branch/val64/inplace/bin/ghc-stage2' -fforce-recomp -dcore-lint -dcmm-lint -dno-debug-output -no-user-package-conf -rtsopts  -fno-ghci-history -c PolyKinds03.hs    >PolyKinds03.comp.stderr 2>&1
Compile failed (status 256) errors were:
ghc-stage2: panic! (the 'impossible' happened)
  (GHC version 7.4.0.20120126 for x86_64-apple-darwin):
        ASSERT failed! file compiler/types/Kind.lhs line 254
ghc-prim:GHC.Prim.BOX{(w) tc 347} ghc-prim:GHC.Prim.ArgKind{(w) tc 34p}

Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

PolyKinds13 fails similarly.

Change History

Changed 16 months ago by igloo

  • testcase set to PolyKinds03 PolyKinds13

Changed 16 months ago by dreixel

  • keywords PolyKinds added

Changed 16 months ago by simonpj

Why is this just on the "slow" validate?

Works on HEAD, and we are not claiming PolyKinds always works on the branch, so I'm not sure this is worth pursuing. Is it? Doubtless we could track down which HEAD patch fixes it...

Simon

Changed 16 months ago by igloo

Oh, presumably it happens with any compiler built with

GhcStage2HcOpts += -DDEBUG

I just wrote "With a slow validate build on OSX" so that I'd know exactly how to reproduce it if necessary in the future.

If the assertion doesn't fire with a DEBUG HEAD, then we can just close the ticket.

Changed 16 months ago by simonpj

I don't get this with DEBUG HEAD, do you?

Changed 16 months ago by igloo

Yes; with a regular validate build, on both OS X and Linux amd64, and

GhcStage2HcOpts += -DDEBUG

in mk/validate.mk I get:

=====> PolyKinds03(normal) 4 of 13 [0, 0, 0]
cd . && '/home/ian/ghc/git/ghc/inplace/bin/ghc-stage2' -fforce-recomp -dcore-lint -dcmm-lint -dno-debug-output -no-user-package-conf -rtsopts  -fno-ghci-history -c PolyKinds03.hs    >PolyKinds03.comp.stderr 2>&1
Compile failed (status 256) errors were:
ghc-stage2: panic! (the 'impossible' happened)
  (GHC version 7.5.20120130 for x86_64-unknown-linux):
        ASSERT failed! file compiler/types/Kind.lhs line 254
ghc-prim:GHC.Prim.BOX{(w) tc 347} ghc-prim:GHC.Prim.ArgKind{(w) tc 34p}

Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

(incidentally, the validate fails with an assertion failure in the vectoriser while building the DPH libs, but stage2 is already built by then)

Changed 16 months ago by simonpj

  • owner set to simonpj

I'm on this

Changed 16 months ago by simonpj

  • status changed from new to closed
  • resolution set to fixed

Fixed by

commit af26d558fb57150a0035cc72d15f829975db2e9f
Author: Simon Peyton Jones <simonpj@microsoft.com>
Date:   Mon Feb 6 08:40:44 2012 +0000

    Give promoted DataCons a tyConArity and promoted TyCons a tyConKind
    
    ..and fix up Core Lint.  (I was getting a bogus Core Lint failure.)

 compiler/basicTypes/DataCon.lhs |   93 +++++++++++++++-
 compiler/coreSyn/CoreLint.lhs   |  231 +++++++++++++++++++++-----------------
 compiler/iface/BuildTyCl.lhs    |    8 --
 compiler/prelude/TysWiredIn.lhs |   10 +-
 compiler/typecheck/TcHsType.lhs |   26 ++---
 compiler/types/Kind.lhs         |   61 +----------
 compiler/types/TyCon.lhs        |   46 ++++----
 7 files changed, 264 insertions(+), 211 deletions(-)
Note: See TracTickets for help on using tickets.