Re: [GHC] #2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669
#2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669 --+- Reporter: batterseapower |Owner: igloo Type: bug| Status: closed Priority: normal |Milestone: 6.10.2 Component: Compiler | Version: 6.9 Severity: normal | Resolution: fixed Keywords: | Difficulty: Unknown Testcase: simplCore/should_compile/simpl014 | Os: MacOS X Architecture: Unknown/Multiple | --+- Changes (by igloo): * status: new = closed * resolution: = fixed Comment: Confirmed: `simpl014` now passes with `./validate --slow`. -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/2368#comment:5 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Re: [GHC] #2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669
#2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669
--+-
Reporter: batterseapower |Owner: igloo
Type: bug| Status: new
Priority: normal |Milestone: 6.10.2
Component: Compiler | Version: 6.9
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: simplCore/should_compile/simpl014 | Os: MacOS X
Architecture: Unknown/Multiple |
--+-
Changes (by simonpj):
* owner: = igloo
Comment:
I believe this is fixed by
{{{
Tue Jan 13 16:49:53 GMT 2009 simo...@microsoft.com
* Rewrite CorePrep and improve eta expansion
}}}
Do not merge to the 6.10 branch (at least not yet); it's a bit radical.
Ian: can you check you agree this is now ok, and close.
Simon
--
Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/2368#comment:4
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Re: [GHC] #2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669
#2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669 ---+ Reporter: batterseapower | Owner: Type: bug| Status: new Priority: normal | Milestone: 6.10.2 Component: Compiler |Version: 6.9 Severity: normal | Resolution: Keywords: | Difficulty: Unknown Testcase: simplCore/should_compile/simpl014 | Architecture: Unknown/Multiple Os: MacOS X| ---+ Changes (by igloo): * milestone: 6.10.1 = 6.10.2 -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/2368#comment:3 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
Re: [GHC] #2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669
#2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669 ---+ Reporter: batterseapower | Owner: Type: bug| Status: new Priority: normal | Milestone: 6.10.1 Component: Compiler |Version: 6.9 Severity: normal | Resolution: Keywords: | Difficulty: Unknown Testcase: simplCore/should_compile/simpl014 | Architecture: Unknown Os: MacOS X| ---+ Changes (by igloo): * difficulty: = Unknown * milestone: = 6.10.1 Comment: This is reproducible with `./validate --slow`. -- Ticket URL: http://hackage.haskell.org/trac/ghc/ticket/2368#comment:1 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler___ Glasgow-haskell-bugs mailing list Glasgow-haskell-bugs@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
[GHC] #2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669
#2368: ASSERT failed! file coreSyn/CorePrep.lhs line 669
---+
Reporter: batterseapower | Owner:
Type: bug | Status: new
Priority: normal | Component: Compiler
Version: 6.9 |Severity: normal
Keywords: |Testcase: simplCore/should_compile/simpl014
Architecture: Unknown | Os: MacOS X
---+
This is exposed by running the testsuite on a stage2 compiler with
-DDEBUG. The tests triggering it are:
* simplCore/should_compile/simpl014
The affected context is:
{{{
etaExpandRhs :: CoreBndr - CoreExpr - UniqSM CoreExpr
etaExpandRhs bndr rhs = do
-- Eta expand to match the arity claimed by the binder
-- Remember, CorePrep must not change arity
--
-- Eta expansion might not have happened already,
-- because it is done by the simplifier only when
-- there at least one lambda already.
--
-- NB1:we could refrain when the RHS is trivial (which can happen
-- for exported things). This would reduce the amount of code
-- generated (a little) and make things a little words for
-- code compiled without -O. The case in point is data
constructor
-- wrappers.
--
-- NB2: we have to be careful that the result of etaExpand doesn't
--invalidate any of the assumptions that CorePrep is
attempting
--to establish. One possible cause is eta expanding inside of
--an SCC note - we're now careful in etaExpand to make sure
the
--SCC is pushed inside any new lambdas that are generated.
--
-- NB3: It's important to do eta expansion, and *then* ANF-ising
-- f = /\a - g (h 3) -- h has arity 2
-- If we ANF first we get
-- f = /\a - let s = h 3 in g s
-- and now eta expansion gives
-- f = /\a - \ y - (let s = h 3 in g s) y
-- which is horrible.
-- Eta expanding first gives
-- f = /\a - \y - let s = h 3 in g s y
--
us - getUniquesM
let eta_rhs = etaExpand arity us rhs (idType bndr)
ASSERT2( manifestArity eta_rhs == arity, (ppr bndr + ppr arity +
ppr (exprArity rhs))
$$ ppr rhs $$ ppr eta_rhs )
-- Assertion checks that eta expansion was successful
return eta_rhs
where
-- For a GlobalId, take the Arity from the Id.
-- It was set in CoreTidy and must not change
-- For all others, just expand at will
arity | isGlobalId bndr = idArity bndr
| otherwise = exprArity rhs
}}}
The precise error is:
{{{
ghc-6.9.20080614: panic! (the 'impossible' happened)
(GHC version 6.9.20080614 for i386-apple-darwin):
ASSERT failed! file coreSyn/CorePrep.lhs line 669
main:ShouldCompile.runHandler{v r6G} [gid] 3 3
__inline_me ((\ (@ st{tv a73} [tv] :: ghc-prim:GHC.Prim.*{(w) tc 34d})
(w{v si9} [lid] :: main:ShouldCompile.M{tc r7}
(main:ShouldCompile.IHR{tc r6} st{tv
a73} [tv]))
(w1{v sia} [lid] :: main:ShouldCompile.IHandler{tc r8}
st{tv a73} [tv])
(eta{v sj4} [lid] :: ghc-prim:GHC.Prim.State#{(w) tc 32q}
ghc-prim:GHC.Prim.RealWorld{(w) tc
31E})
(eta1{v sjb} [lid] :: ghc-prim:GHC.Prim.State#{(w) tc 32q}
ghc-prim:GHC.Prim.RealWorld{(w) tc
31E}) -
main:ShouldCompile.$wrunHandler{v r2} [gid]
@ st{tv a73} [tv] w1{v sia} [lid] eta1{v sjb} [lid])
`cast` (ghc-prim:GHC.Prim.trans{(w) tc 34y}
(forall st{tv a73} [tv].
main:ShouldCompile.M{tc r7}
(main:ShouldCompile.IHR{tc r6} st{tv a73} [tv])
- main:ShouldCompile.IHandler{tc r8} st{tv a73}
[tv]
- ghc-prim:GHC.Prim.State#{(w) tc 32q}
ghc-prim:GHC.Prim.RealWorld{(w) tc 31E}
- ghc-prim:GHC.Prim.CoUnsafe{(w) tc 34K}
(ghc-prim:GHC.Prim.State#{(w) tc 32q}
ghc-prim:GHC.Prim.RealWorld{(w) tc 31E}
- (# ghc-prim:GHC.Prim.State#{(w) tc 32q}
ghc-prim:GHC.Prim.RealWorld{(w) tc
31E},
base:GHC.Base.(){(w) tc 40} #))
