GADTs with newtype deriving can crash GHCi / cause Core Lint to fail

In my attempts to prove that GHC was generating errors on valid code, I have stumbled on a way to crash six versions of GHC, both F and FC, and in one case passing Core Lint!

stefan@stefans:/tmp$ V=6.4.2 ; ghc-$V -V ; ghc-$V -e 'seq icA ()' A.hs ; ghc-$V -dcore-lint -e 'seq icA ()' A.hs
The Glorious Glasgow Haskell Compilation System, version 6.4.2
*** Exception: Segmentation fault
*** Exception: Segmentation fault
stefan@stefans:/tmp$ V=6.6 ; ghc-$V -V ; ghc-$V -e 'seq icA ()' A.hs ; ghc-$V -dcore-lint -e 'seq icA ()' A.hs
The Glorious Glasgow Haskell Compilation System, version 6.6
mkCase1: null alts tpl_X6{v} [lid] main:A.icA{v rFP} [gid]
Segmentation fault
mkCase1: null alts tpl_X6{v} [lid] main:A.icA{v rFP} [gid]
ghc-6.6: panic! (the 'impossible' happened)
  (GHC version 6.6 for i386-unknown-linux):
        compileExpr
    <no location info>:
        In the expression: base:GHC.Err.error{(w) v 06} [gid]
                             @ base:GHC.Base.(){(w) tc 40} "Impossible alternative"
        Argument value doesn't match argument type:
        Fun type:
            [base:GHC.Base.Char{(w) tc 3o}] -> base:GHC.Base.(){(w) tc 40}
        Arg type: base:GHC.Prim.Addr#{(w) tc 33}
        Arg: "Impossible alternative"

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

stefan@stefans:/tmp$ V=6.7 ; ghc-$V -V ; ghc-$V -e 'seq icA ()' A.hs ; ghc-$V -dcore-lint -e 'seq icA ()' A.hs
The Glorious Glasgow Haskell Compilation System, version 6.7
mkCase1: null alts tpl_B1{v} [lid] main:A.icA{v rq2} [gid]
Segmentation fault
mkCase1: null alts tpl_B1{v} [lid] main:A.icA{v rq2} [gid]
ghc-6.7: panic! (the 'impossible' happened)
  (GHC version 6.7 for i386-unknown-linux):
        compileExpr
    <no location info>:
        In the expression: base:GHC.Err.error{(w) v 06} [gid]
                             @ base:GHC.Base.(){(w) tc 40} "Impossible alternative"
        Argument value doesn't match argument type:
        Fun type:
            [base:GHC.Base.Char{(w) tc 3o}] -> base:GHC.Base.(){(w) tc 40}
        Arg type: base:GHC.Prim.Addr#{(w) tc 33}
        Arg: "Impossible alternative"

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

stefan@stefans:/tmp$ V=6.7.20070213 ; ghc-$V -V ; ghc-$V -e 'seq icA ()' A.hs ; ghc-$V -dcore-lint -e 'seq icA ()' A.hs
The Glorious Glasgow Haskell Compilation System, version 6.7.20070213
mkCase: null alts tpl_B1{v} [lid] main:A.icA{v rsd} [gid]
Segmentation fault
mkCase: null alts tpl_B1{v} [lid] main:A.icA{v rsd} [gid]
ghc-6.7.20070213: panic! (the 'impossible' happened)
  (GHC version 6.7.20070213 for i386-unknown-linux):
        compileExpr
    <no location info>:
        In the expression: base:GHC.Err.error{(w) v 06} [gid]
                             @ base:GHC.Base.(){(w) tc 40} "Impossible alternative"
        Argument value doesn't match argument type:
        Fun type:
            [base:GHC.Base.Char{(w) tc 3o}] -> base:GHC.Base.(){(w) tc 40}
        Arg type: base:GHC.Prim.Addr#{(w) tc 33}
        Arg: "Impossible alternative"

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

stefan@stefans:/tmp$ V=6.7.20070223 ; ghc-$V -V ; ghc-$V -e 'seq icA ()' A.hs ; ghc-$V -dcore-lint -e 'seq icA ()' A.hs
The Glorious Glasgow Haskell Compilation System, version 6.7.20070223
mkCase: null alts tpl_B1{v} [lid] main:A.icA{v rs9} [gid]
Segmentation fault
mkCase: null alts tpl_B1{v} [lid] main:A.icA{v rs9} [gid]
ghc-6.7.20070223: panic! (the 'impossible' happened)
  (GHC version 6.7.20070223 for i386-unknown-linux):
        compileExpr
    <no location info>:
        In the expression: base:GHC.Err.error{(w) v 06} [gid]
                             @ base:GHC.Base.(){(w) tc 40} "Impossible alternative"
        Argument value doesn't match argument type:
        Fun type:
            [base:GHC.Base.Char{(w) tc 3o}] -> base:GHC.Base.(){(w) tc 40}
        Arg type: base:GHC.Prim.Addr#{(w) tc 33}
        Arg: "Impossible alternative"

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

stefan@stefans:/tmp$ V=6.7.20070323 ; ghc-$V -V ; ghc-$V -e 'seq icA ()' A.hs ; ghc-$V -dcore-lint -e 'seq icA ()' A.hs
The Glorious Glasgow Haskell Compilation System, version 6.7.20070323
mkCase: null alts tpl_B1{v} [lid] main:A.icA{v rs4} [gid]
Segmentation fault
mkCase: null alts tpl_B1{v} [lid] main:A.icA{v rs4} [gid]
ghc-6.7.20070323: panic! (the 'impossible' happened)
  (GHC version 6.7.20070323 for i386-unknown-linux):
        compileExpr
    <no location info>:
        In the expression: base:GHC.Err.error{(w) v 06} [gid]
                             @ base:GHC.Base.(){(w) tc 40} "Impossible alternative"
        Argument value doesn't match argument type:
        Fun type:
            [base:GHC.Base.Char{(w) tc 3o}] -> base:GHC.Base.(){(w) tc 40}
        Arg type: base:GHC.Prim.Addr#{(w) tc 33}
        Arg: "Impossible alternative"

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

stefan@stefans:/tmp$ cat A.hs
{-# OPTIONS_GHC -fglasgow-exts #-}
module A(II(..), icA) where

data II a where II :: II Int

class IC a where ic :: II a

instance IC Int where ic = II

newtype A = A Int deriving(IC)

icA = (ic :: II A)

Trac metadata

Trac field	Value
Version	6.6
Type	Bug
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	Compiler
Test case
Differential revisions
BlockedBy
Related
Blocking
CC
Operating system
Architecture