ghc: panic! (the 'impossible' happened) - piResultTy

Oddly this only happens on our build server, not on my local machine. On compilation, I get:

[187 of 273] Compiling Query.Order      ( Query/Order.hs, dist/build/Query/Order.o )

Query/Order.hs:340:1: warning: [-Wredundant-constraints]
    • Redundant constraint: MonadBaseControl IO m
    • In the type signature for:
           updateOrdersSuccess :: (MonadBaseControl IO m,
                                   MonadTransaction m) =>
                                  m ()
ghc: panic! (the 'impossible' happened)
  (GHC version 8.0.1 for x86_64-unknown-linux):
	piResultTy
  Maybe Int64
  a1_a4WiL

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

The configuration parameters are:

configureFlags: --verbose --prefix=/nix/store/qxfjd5jxr593dn43i3659s86d0rdllyv-circuithub-api-0.0.4 --libdir=$prefix/lib/$compiler --libsubdir=$pkgid --with-gcc=gcc --package-db=/tmp/nix-build-circuithub-api-0.0.4.drv-0/package.conf.d --ghc-option=-j1 --disable-split-objs --disable-library-profiling --disable-profiling --disable-shared --enable-library-vanilla --disable-executable-dynamic --disable-tests

I've attached Query/Order.hs, but you won't be able to compile it stand-alone.

Trac metadata

Trac field	Value
Version	8.0.1
Type	Bug
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	Compiler
Test case
Differential revisions
BlockedBy
Related
Blocking
CC
Operating system
Architecture

changed weight to 5

added Tbug Trac import labels

Attached file Order.hs (download).

Query.Order

This is a definite bug. It'll be hard to do much about this without the ability to reproduce the bug.

Could you compile everything with -dore-lint?

Maybe it's already fixed. cf #12676 (closed) Any chance you can try with HEAD?

Sadly we're still seeing this on GHC 8.0.2 RC 1. I'll try and compile with -dcore-lint now and see if that shines any light on the issue.

Trac metadata

Trac field	Value
Version	8.0.1 → 8.0.2-rc1

It's be great to nail this but we are stuck without a way to reproduce it.

Don't worry, perfectly understandable! I'm going to try and get some more information on this for you. I can't even reliably reproduce it myself, some builds are going through while others fail - always on this module though. I'm hoping -dcore-lint will give us a bit more understanding of what is going wrong though.

We've had a few more panics today, each different:

[248 of 278] Compiling Handler.Auth     ( src/Handler/Auth.hs, dist/build/Handler/Auth.o )
ghc: panic! (the 'impossible' happened)
  (GHC version 8.0.1.20161117 for x86_64-unknown-linux):
	Iface Lint failure
  In interface for Data.ByteString.Internal
  Unfolding of $fShowByteString_$cshow
    <no location info>: warning:
        In the expression: $trModule2
                             (unpackChars x_apbxX) (: @ Char shows6 ([] @ Char))
        Non-function type in function position
        Fun type: TrName
        Arg type: [Char]
        Arg: unpackChars x_apbxX
  $fShowByteString_$cshow = \ (x_apbxX :: ByteString) ->
                              : @ Char
                                shows6
                                ($trModule2 (unpackChars x_apbxX) (: @ Char shows6 ([] @ Char)))
  Iface expr = \ (x :: ByteString) ->
               : @ Char
                 shows6
                 (showLitString (unpackChars x) (: @ Char shows6 ([] @ Char)))

and

[215 of 278] Compiling Foundation       ( Foundation.hs, dist/build/Foundation.o )
ghc: panic! (the 'impossible' happened)
  (GHC version 8.0.1.20161117 for x86_64-unknown-linux):
	tyThingTyCon Identifier ‘$fEnumIsolationLevel’

I'll continue to see if I can get any sort of repro on this.

Another rather huge core lint offense this time:

https://gist.github.com/ocharles/5237072c9a5620198e31d74baab388b8

Another panic that is different from all the above:

[231 of 278] Compiling Handler.Pallets  ( src/Handler/Pallets.hs, dist/build/Handler/Pallets.o )
ghc: panic! (the 'impossible' happened)
  (GHC version 8.0.1.20161117 for x86_64-unknown-linux):
	getClassPredTys [Sql]

It's very strange that we seem to get different panics every time

All very mysterious. But I can't do anything without a way to reproduce, unfortunately.

Did this turn out to be the issue with uniques overflowing? Or was that something else?

closed

I think so - haven't seen a panic since! Let's go ahead and close this

Trac metadata

Trac field	Value
Resolution	Unresolved → ResolvedFixed

Great. (Wanted to make sure I wasn't confusing this with a different issue.)

added Pnormal label