Hi Yamamoto-san,

I've encountered the same problem before, and there's a haddock issue tracking this problem: https://github.com/haskell/haddock/issues/837. A simple fix is building latest version of haddock and replacing the haddock binary for ghc-8.4.3.

On Fri, Jul 20, 2018 at 4:29 PM, Yuji Yamamoto <whosekiteneverfly@gmail.com> wrote:
Thank you Michael and Vanessa!

Here're what I've tried and their results:

A. `cabal new-haddock` in a project depending on haskell-src-exts as you suggested.
    Resulted in another, non-related error..:
    > cabal new-haddock --with-ghc='C:\Users\yuji-yamamoto\AppData\Local\Programs\stack\x86_64-windows\ghc-8.4.3\bin\ghc.exe'
    Resolving dependencies...
    Build profile: -w ghc-8.4.3 -O1
    In order, the following will be built (use -v for more details):
     - happy-1.19.9 (exe:happy) (requires build)
     - haskell-src-exts-1.20.2 (lib) (requires build)
     - proceedings-bot-0.1.0.0 (lib) (first run)
    Configuring happy-1.19.9 (all, legacy fallback)...
    cabal.exe: Failed to build happy-1.19.9 (which is required by
    proceedings-bot-0.1.0.0). The failure occurred during the configure step. The
    exception was:
    output of
    C:\Users\yuji-yamamoto\AppData\Local\Programs\stack\x86_64-windows\ghc-8.4.3\bin\ghc.exe:
    hGetContents: invalid argument (invalid byte sequence)

B. Running a command extracted a log from `stack haddock --verbose haskell-src-exts`.
    Resulted in the same error
    >  C:\sr\setup-exe-cache\x86_64-windows\Cabal-simple_Z6RU0evB_2.2.0.1_ghc-8.4.3.exe --builddir=.stack-work\dist\7d103d30 haddock --html --hoogle  --haddock-option=--hyperlinked-source --verbose
    ...
    Access violation in generated code when reading 0x36f00000

     Attempting to reconstruct a stack trace...

       Frame        Code address
     * 0x440daf0    0xe63d37 C:\Users\yuji-yamamoto\AppData\Local\Programs\stack\x86_64-windows\ghc-8.4.3\bin\haddock.exe+0xa63d37
     * 0x440daf8    0x7e1100049f40 C:\Users\yuji-yamamoto\p\cmder_mini\vendor\conemu-maximus5\ConEmu\ConEmuHk64.dll+0x49f40

C. Plus I found `stack haddock` after `stack unpack haskell-src-exts-1.20.2 && cd  haskell-src-exts-1.20.2
     Sounds weird...

I'm happy if getting some more advice to make smaller case.

Thanks,

    

2018年7月19日(木) 22:35 Michael Snoyman <michael@snoyman.com>:
Given that the haddock.exe executable itself is crashing, it seems highly unlikely to be a bug in either Stack or Cabal. I can think of three possibilities here:

* There's a bug in Haddock itself, such as an FFI call or some embedded C code
* There's a bug in GHC which is generating invalid code
* The GHC bindist itself has included a corrupted haddock.exe

I agree with Vanessa: trying to reproduce this with another build tool (`cabal haddock` or `cabal new-haddock`) would be a good next step, followed by a bug report against (most likely) GHC.

On Thu, Jul 19, 2018 at 9:29 AM Yuji Yamamoto <whosekiteneverfly@gmail.com> wrote:
I always get an " Access violation in generated code when reading 0x28900000" error when
executing

> stack haddock haskell-src-exts

since LTS 12.0.

I'm not sure to which project I should submit a bug report.
GHC? Haddock? Or Stack?
And how can I make a smaller case  to reproduce the same error?

I  attached a log when executing with --verbose --cabal-verbose for your info.

--
_______________________________________________
Haskell-Cafe mailing list
To (un)subscribe, modify options or view archives go to:
http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
Only members subscribed via the mailman list are allowed to post.


--

_______________________________________________
Haskell-Cafe mailing list
To (un)subscribe, modify options or view archives go to:
http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
Only members subscribed via the mailman list are allowed to post.