internal error: m32_allocator_init: Failed to map

Are you running in an environment with an address space ulimit, by any chance?

added Phigh RTS labels and removed needs triage label

Not that I can see:

$ ulimit -a
core file size          (blocks, -c) 0
data seg size           (kbytes, -d) unlimited
scheduling priority             (-e) 0
file size               (blocks, -f) unlimited
pending signals                 (-i) 27673
max locked memory       (kbytes, -l) 65536
max memory size         (kbytes, -m) unlimited
open files                      (-n) 65536
pipe size            (512 bytes, -p) 8
POSIX message queues     (bytes, -q) 819200
real-time priority              (-r) 0
stack size              (kbytes, -s) 8192
cpu time               (seconds, -t) unlimited
max user processes              (-u) 27673
virtual memory          (kbytes, -v) unlimited
file locks                      (-x) unlimited

Interesting, it works fine with ghc-8.6.5

On Mac there is also a failure, but it is very different:

doctests: 
lookupSymbol failed in relocateSection (relocate external)
/Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a: unknown symbol `__ZNK17double_conversion23DoubleToStringConverter11ToPrecisionEdiPNS_13StringBuilderE'
GHC runtime linker: fatal error: I found a duplicate definition for symbol
   __ZNK17double_conversion6VectorIcEixEi
whilst processing object file
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a
The symbol was previously defined in
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a(hs-double-conversion.o)
This could be caused by:
   * Loading two different object files which export the same symbol
   * Specifying the same object file twice on the GHCi command line
   * An incorrect `package.conf' entry, causing some object to be
     loaded twice.
doctests: 
lookupSymbol failed in relocateSection (relocate external)
/Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a: unknown symbol `__ZN17double_conversion6StrtofENS_6VectorIKcEEi'
GHC runtime linker: fatal error: I found a duplicate definition for symbol
   __ZN17double_conversion7BitCastIdyEET_RKT0_
whilst processing object file
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a
The symbol was previously defined in
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a(double-conversion.o)
This could be caused by:
   * Loading two different object files which export the same symbol
   * Specifying the same object file twice on the GHCi command line
   * An incorrect `package.conf' entry, causing some object to be
     loaded twice.
GHC runtime linker: fatal error: I found a duplicate definition for symbol
   __ZN17double_conversion6Bignum9LessEqualERKS0_S2_
whilst processing object file
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a
The symbol was previously defined in
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a(bignum-dtoa.o)
This could be caused by:
   * Loading two different object files which export the same symbol
   * Specifying the same object file twice on the GHCi command line
   * An incorrect `package.conf' entry, causing some object to be
     loaded twice.
GHC runtime linker: fatal error: I found a duplicate definition for symbol
   __ZN17double_conversion7BitCastIdyEET_RKT0_
whilst processing object file
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a
The symbol was previously defined in
   /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a(double-conversion.o)
This could be caused by:
   * Loading two different object files which export the same symbol
   * Specifying the same object file twice on the GHCi command line
   * An incorrect `package.conf' entry, causing some object to be
     loaded twice.
doctests: Could not load Object Code /Users/runner/.cabal/store/ghc-8.10.4/dbl-cnvrsn-2.0.2.0-2cca6009/lib/libHSdbl-cnvrsn-2.0.2.0-2cca6009.a(fast-dtoa.o).

doctests: doctests: unable to load package `double-conversion-2.0.2.0'

I have also raised a ticket against doctest: https://github.com/sol/doctest/issues/293

@newhoggy, is this reproducible with %8.10.4? There were a good number of linker fixes in this release..

I have tried to reproduce this with d35b2d629e44649927083a395f00bc3721a39621 (since the cited commit appears to be no longer available). Unfortunately, it appears not to build, failing with overlapping instances between ListLike and utf8-string:

[13 of 21] Compiling Data.ListLike.UTF8 ( src/Data/ListLike/UTF8.hs, dist/build/Data/ListLike/UTF8.o, dist/build/Data/ListLike/UTF8.dyn_o )

src/Data/ListLike/UTF8.hs:153:10: error:
    • Overlapping instances for IsString (UTF8 BSC.ByteString)
        arising from the superclasses of an instance declaration
      Matching instances:
        instance UTF8.UTF8Bytes string index => IsString (UTF8 string)
          -- Defined in ‘Data.String.UTF8’
        instance IsString (UTF8 BSC.ByteString)
          -- Defined at src/Data/ListLike/UTF8.hs:150:10
    • In the instance declaration for
        ‘StringLike (UTF8 BSC.ByteString)’
    |
153 | instance StringLike (UTF8 BS.ByteString) where
    |          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

src/Data/ListLike/UTF8.hs:277:10: error:
    • Overlapping instances for IsString (UTF8 BSLC.ByteString)
        arising from the superclasses of an instance declaration
      Matching instances:
        instance UTF8.UTF8Bytes string index => IsString (UTF8 string)
          -- Defined in ‘Data.String.UTF8’
        instance IsString (UTF8 BSLC.ByteString)
          -- Defined at src/Data/ListLike/UTF8.hs:274:10
    • In the instance declaration for
        ‘StringLike (UTF8 BSLC.ByteString)’
    |
277 | instance StringLike (UTF8 BSL.ByteString) where
    |          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

@newhoggy, any progress here?

added info needed label

I've switched to use cabal-docspec to avoid the problem. The last commit before I did that was:

https://github.com/input-output-hk/cardano-ledger-specs/commit/d35b2d629e44649927083a395f00bc3721a39621

I've created a build here. Hopefully it reproduces there. https://github.com/input-output-hk/cardano-ledger-specs/pull/2199

Okay, I've reproduced it in this build: https://github.com/input-output-hk/cardano-ledger-specs/pull/2199/checks?check_run_id=2197293721

Here is another instance of this: https://github.com/haskell-hint/hint/issues/131 (using hint, servant, hspec)

@jwaldmann how easily can you reproduce your case on-demand?

I suspect we will need to add further debug output to this failure message to get any traction on debugging it. This sort of error can arise from many failure modes and without knowing more about the state of the address space it is very hard to say which is responsible.

Hi. I now spent some hours on this (*) , and give up because of a Heisenbug feeling. E.g., the behaviour changes after stack clean, and stack test is different from running the test from stack exec -- ghci .... So, I'm not sure whether I am debugging ghci, cabal, stack, or hint.

For my application, it seems that stack test does work in our CI setting, and that's enough for the moment.

(*) part of a larger repo, but this subdir is stand-alone: https://gitlab.imn.htwk-leipzig.de/autotool/all0/-/tree/master/hint-check

Very unfortunate; thanks for trying, @jwaldmann!

@bgamari, I can reproduce this bug 100% cases when use Haskell interpreter with hint and ghc-8.10.4 on Linux, though there is no bug on MacOS BigSur. The bug appears after nixpkgs channel upgrade from release-20.09 to release-21.05. If I set -xm with something bug disappears (e.g. -xm50 or -xm100). I realized, I cannot set default -xm with -with-rtsopts.

uname -a
Linux diehard 5.9.13 #1 SMP Wed Dec 9 14:12:51 MSK 2020 x86_64 GNU/Linux

Script being evaluated by hint:

module Hello where
pureStr :: String
pureStr = "Hello World"

Hint args:

["-hide-all-packages","-no-user-package-db","-package-db","/home/dan/pro/app/dist-newstyle/app-package-db","-package-db","/nix/store/v2jfgbbi3ciwql8bp4p7cpcl1prg3dl0-ghc-8.10.4-with-packages/lib/ghc-8.10.4/package.conf.d","-package","app"]

Hint imports:

import qualified Language.Haskell.Interpreter as HI
import qualified Language.Haskell.Interpreter.Unsafe as HI

      HI.setImports [ "App.Types" ]

This is a rather tricky bug since it's dependent upon how the operating system decides to layout our requested memory mappings. For what it's worth, I merged quite a few cleanups of the m32 allocator in 9.0 (!4497 (closed), !2033 (closed), !5567 (closed)) which aren't present in %8.10.4 (or any 8.10, for that matter). Perhaps it would be worth backporting these.

added 1 deleted label

internal error: m32_allocator_init: Failed to map

Summary

Steps to reproduce

Expected behavior

Environment

Child items ...

Activity