Panic: collectNBinders

changed weight to 5

changed the description

Note: also reproduces with GHC 8.2.2 (stackage lts-11.7).

$ stack ghc -- --version
The Glorious Glasgow Haskell Compilation System, version 8.2.2

$ stack build --profile
bug-0.1.0.0: configure (exe)
Configuring bug-0.1.0.0...
bug-0.1.0.0: build (exe)
Preprocessing executable 'bug' for bug-0.1.0.0..
Building executable 'bug' for bug-0.1.0.0..
[1 of 1] Compiling Main             ( Main.hs, .stack-work/dist/x86_64-linux-tinfo6-nopie/Cabal-2.0.1.0/build/bug/bug-tmp/Main.p_o )
ghc: panic! (the 'impossible' happened)
  (GHC version 8.2.2 for x86_64-unknown-linux):
	collectNBinders
  1
  Call stack:
      CallStack (from HasCallStack):
        prettyCurrentCallStack, called at compiler/utils/Outputable.hs:1133:58 in ghc:Outputable
        callStackDoc, called at compiler/utils/Outputable.hs:1137:37 in ghc:Outputable
        pprPanic, called at compiler/coreSyn/CoreSyn.hs:1997:25 in ghc:CoreSyn

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


--  While building custom Setup.hs for package bug-0.1.0.0 using:
      /usr/local/home/cdisselk/.stack/setup-exe-cache/x86_64-linux-tinfo6-nopie/Cabal-simple_mPHDZzAJ_2.0.1.0_ghc-8.2.2 --builddir=.stack-work/dist/x86_64-linux-tinfo6-nopie/Cabal-2.0.1.0 build exe:bug --ghc-options " -ddump-hi -ddump-to-file -fdiagnostics-color=always"
    Process exited with code: ExitFailure 1

mentioned in commit 4e45ebee

Thanks for the nice test case, I added it to the test suite.

JFTR, this can be easily reproduced without stack or GHC:

~/build/haskell/ghc-master $ ./inplace/bin/ghc-stage2  testsuite/tests/profiling/should_compile/T15108.hs  -prof -fprof-auto -dcore-lint
[1 of 1] Compiling Main             ( testsuite/tests/profiling/should_compile/T15108.hs, testsuite/tests/profiling/should_compile/T15108.o )
*** Core Lint errors : in result of Desugar (after optimization) ***
testsuite/tests/profiling/should_compile/T15108.hs:22:7: warning:
    [RHS of myInt_aVZ :: Int -> MyInt]
    Join point has too few lambdas
    Join var: myInt_aVZ
    Join arity: 1
    Number of lambdas: 0
    Rhs =  scctick<getInt.myInt> MyInt
*** Offending Program ***
getInt :: Shape -> MyInt
[LclIdX]
getIntghc-stage2: panic! (the 'impossible' happened)
  (GHC version 8.5.20180410 for x86_64-unknown-linux):
	collectNBinders
  1
  Call stack:
      CallStack (from HasCallStack):
        callStackDoc, called at compiler/utils/Outputable.hs:1162:37 in ghc:Outputable
        pprPanic, called at compiler/coreSyn/CoreSyn.hs:2056:25 in ghc:CoreSyn

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

As you can see, it does not require any optimization level, and the bug is in place right after the desugarer.

Maybe the optimizer copies the join arity from MyInt to myInt, under the assumption that myInt will be inlined, but the scctick prevents that from happening?

I'm on this.

mentioned in commit 07cc6039

mentioned in commit d4cc74f1

closed

changed milestone to %8.6.1

OK, fixed. Thanks for a great report.

Trac metadata

Trac field	Value
Resolution	Unresolved → ResolvedFixed

Have you considered adding

  | JoinNonRec v [v] (Expr v)
  | JoinRec [(v, [v], Expr v)]

to Bind v? It seems that a great lot of code has to zoom past the binders of a join point, causing bugs. With these constructors, one would be forced to handle them properly – and code that does not need to pay attention can use a pattern synonym that recreates the original view.

I suppose we could, but a key aspect of the current join points is that their semantics is exactly that of let-bindings; and often code (e.g. strictness analysis) can work uniformly for both.

Yes, there are many ways in which we could change the representation. E.g. maybe Lam should take a list of binders, or App should take a list of arguments. But my instinct is that you save code in some places and lose it in others.

Yeah, it’d probably amount to shifting boxes around without making any new space :-)

added compiler crash label

added Pnormal label

mentioned in issue #23922 (closed)

Panic: collectNBinders

Child items ...

Activity