#1654: GHC Panic in 6.8.1
-----------------------------+----------------------------------------------
Reporter: [EMAIL PROTECTED] | Owner:
Type: bug | Status: new
Priority: high | Milestone: 6.8 branch
Component: Compiler | Version: 6.8.1
Severity: major | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: x86
Os: Linux |
-----------------------------+----------------------------------------------
Comment (by simonpj):
OK, so here's what I've seen so far. In `ghc-log.bz2` I see
* That there is a Core Lint failure in the stage 2 build, for file
`RnTypes.lhs`.
* Stage 2 is built with `-O2`
* The Core Lint failure is in the result of `SpecConstr`
* In the specialisation of a join point in `mk_hs_op_ty`.
There is something odd going on, because there are duplicate
specialisations for this join point:
{{{
$j_X6vX [Once*! Nothing] :: GHC.Base.Bool
-> GHC.Base.Bool
-> IOEnv.IOEnv
(TcRnTypes.Env TcRnTypes.TcGblEnv
TcRnTypes.TcLclEnv)
(HsTypes.HsType Name.Name)
[Arity 2
Str: DmdType SL
RULES: "SC:$j_X6vX0" [0] forall {} $j_X6vX GHC.Base.False GHC.Base.True
= $s$j_s6Qh
"SC:$j_X6vX1" [0] forall {} $j_X6vX GHC.Base.True GHC.Base.False
= $s$j_s6Qi
"SC:$j_X6vX2" [0] forall {} $j_X6vX GHC.Base.False
GHC.Base.False = $s$j_s6Qj
"SC:$j_X6vX0" [0] forall {} $j_X6vX GHC.Base.False GHC.Base.True
= $s$j_s6PF
"SC:$j_X6vX1" [0] forall {} $j_X6vX GHC.Base.True GHC.Base.False
= $s$j_s6PG
"SC:$j_X6vX2" [0] forall {} $j_X6vX GHC.Base.False
GHC.Base.False = $s$j_s6PH]
}}}
There should be three specialisations but there are six. The Core Lint
failure is inside a couple of the specialised functions `$s$j_s6PF` for
example.
I have tried to reproduce this odd specialisation behaviour in 6.8.1 and
the HEAD (by compiling `RnTypes` with `-O2`, but both generate three
specialisations (and no Lint failure) as they should.
So let's see if anyone can reproduce this with 6.8.1 or the HEAD. It's
entirely possible that it's fixed already, but it'd be more reassuring to
track it down and thereby be certain.
Simon
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/1654#comment:7>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
