#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.4.3 Component: Compiler | Version: 8.4.2-rc1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Changes (by sjakobi): * cc: sjakobi (added) -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:1 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.4.3 Component: Compiler | Version: 8.4.2-rc1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by sjakobi): I can reproduce the panic when compiling the attached file with `-O1` or `-O2`. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:2 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.4.3 Component: Compiler | Version: 8.4.2-rc1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by sjakobi): I cannot reproduce the panic with ghc-HEAD (`c2f90c84e`) or ghc v8.5.20180306 from hvr's PPA. It is interesting (but possibly unrelated) that the testsuite for `order- maintenance` fails when built with ghc-HEAD. With ghc-8.2.2 it passes. {{{ ~/tmp/order-maintenance-0.2.1.0 $ cabal new-test -w ~/src/ghc/inplace/bin /ghc-stage2 -O1 --allow-newer Build profile: -w ghc-8.5.20180402 -O1 In order, the following will be built (use -v for more details): - order-maintenance-0.2.1.0 (test:tests) (first run) Preprocessing test suite 'tests' for order-maintenance-0.2.1.0.. Building test suite 'tests' for order-maintenance-0.2.1.0.. [1 of 1] Compiling Main ( /home/simon/tmp/order- maintenance-0.2.1.0/dist-newstyle/build/x86_64-linux/ghc-8.5.20180402 /order-maintenance-0.2.1.0/t/tests/build/testsStub/testsStub- tmp/testsStub.hs, /home/simon/tmp/order-maintenance-0.2.1.0/dist- newstyle/build/x86_64-linux/ghc-8.5.20180402/order- maintenance-0.2.1.0/t/tests/build/testsStub/testsStub-tmp/Main.o ) Linking /home/simon/tmp/order-maintenance-0.2.1.0/dist- newstyle/build/x86_64-linux/ghc-8.5.20180402/order- maintenance-0.2.1.0/t/tests/build/testsStub/testsStub ... Running 1 test suites... Test suite tests: RUNNING... Test case Dumb vs. Dietz and Sleator O(log n) amortized time: Fail "(after 14 tests and 5 shrinks):Exception: Map.!: given key is not an element in the map CallStack (from HasCallStack): error, called at libraries/containers/Data/Map/Internal.hs:610:17 in containers-0.5.11.0:Data.Map.InternalnewMinimum -> x_1, newAfter x_1 -> x_2, newMinimum -> x_3, newBefore x_2 -> x_4, newAfter x_4 -> x_5, delete x_1, newBefore x_1 -> x_7" Test suite tests: FAIL ... }}} -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:4 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.4.3 Component: Compiler | Version: 8.4.2-rc1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by simonpj): The bug is in this line of `Exitify` {{{ -- The possible arguments of this exit join point args = filter (`elemVarSet` fvs) captured }}} You have to do at least `map zapStableUnfolding` over args. Consider {{{ joinrec j x = let {-# INLINE foo #-} foo y = blah in case x of True -> j2 foo (h x) False -> ...j.... }}} We are going to exitify the True branch of the 'case x'. But 'foo' is captured, so we get {{{ join exit foo x = jump j2 foo (h x) in joinrec j x = let {-# INLINE foo #-} foo y = blah in case x of True -> jump exit foo x False -> ...j.... }}} But we must zap `foo`'s unfolding when we lambda-abstract it! In effect, the lambda-bound `foo` is an entirely new `foo`, unrelated to the original. Actually, arguably we should give the lambda-bound foo `vanillaIdInfo`. We do this in `SetLevels.abstractVars`. I think that'd be safer. We don't seem to have a function `zapAllIdInfo` but we probably should. Over to you. Please include some version of this explanation, of course. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:6 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.4.2 Component: Compiler | Version: 8.4.2-rc1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Changes (by RyanGlScott): * milestone: 8.4.3 => 8.4.2 Comment: I'm a bit confused as to what the current status of this ticket is. comment:5 identifies a commit (not yet merged into GHC 8.4.2) which apparently fixes this issue, but simonpj suggests in comment:6 that there is more to be done. Either way, we should backport //some// sort of fix to 8.4.2. Thoughts, noneata? -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:8 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: 8.4.2 Component: Compiler | Version: 8.4.2-rc1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by Joachim Breitner ): In [changeset:"3f59d3802170f495702674b4f8e4e80247803654/ghc" 3f59d380/ghc]: {{{ #!CommitTicketReference repository="ghc" revision="3f59d3802170f495702674b4f8e4e80247803654" Add test case for #15005 this succeeds on `master` right now, but I confirmed that it fails on ghc-8.4.1-release. }}} -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:10 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: patch Priority: normal | Milestone: 8.4.2 Component: Compiler | Version: 8.4.2-rc1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Phab:D4582 Wiki Page: | -------------------------------------+------------------------------------- Comment (by Joachim Breitner ): In [changeset:"3cfb12d8adac37e5565d66fd173e4648cc041e65/ghc" 3cfb12d/ghc]: {{{ #!CommitTicketReference repository="ghc" revision="3cfb12d8adac37e5565d66fd173e4648cc041e65" In Exitify, zap idInfo of abstracted variables (fixes #15005) as helpfully outlined by SPJ. This commit copies a small bit code from `SetLevels` which could reasonably be put in `Id` as `zapAllIdinfo`; I did not do this to make merging this commmit into `ghc-8.4` easier. If this commit gets merged, then presumably after commit 3f59d3802170f495702674b4f8e4e80247803654 (test case) and ae0cff0a1834d8b041b06d0e1ab6ce969aac44c8 (other fixes to Exitify.hs). Differential Revision: https://phabricator.haskell.org/D4582 }}} -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:12 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler

#15005: GHC fails with “StgCmmEnv: variable not found” when trying to compile order-maintenance-0.2.1.0 -------------------------------------+------------------------------------- Reporter: jeltsch | Owner: (none) Type: bug | Status: closed Priority: normal | Milestone: 8.4.2 Component: Compiler | Version: 8.4.2-rc1 Resolution: fixed | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Compile-time | Unknown/Multiple crash or panic | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Phab:D4582 Wiki Page: | -------------------------------------+------------------------------------- Changes (by bgamari): * status: merge => closed * resolution: => fixed Comment: Merged to `ghc-8.4` as 2f5b97c37dbc9708e7a33b4e36f66a12b32ee09f. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15005#comment:14 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler