GHC panic!

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

GHC panic!

北原真一
I am asking for help I am Haskeller of Japan I'm using GHC7.6.3 GHC has caused the panic

_______________________________________________
Haskell-Cafe mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/haskell-cafe
Reply | Threaded
Open this post in threaded view
|

Re: GHC panic!

Mateusz Kowalczyk
On 08/16/2014 08:41 AM, 北原真一 wrote:
> I am asking for help I am Haskeller of Japan I'm using GHC7.6.3 GHC has
> caused the panic
> https://gist.github.com/minamiyama1994/6697c3fb9c345da6715f
>
>

It works for me with GHC 7.8.3:

{-
[nix-shell:/tmp]$ ghc M.hs -o M



[1 of 1] Compiling Main             ( M.hs, M.o )

==================== Grand total simplifier statistics ====================
Total ticks:     30

9 PreInlineUnconditionally
  4 tpl_B1
  4 tpl_B2
  1 ds_db1Z
4 UnfoldingDone 4 GHC.Base.$
17 BetaReduction
  4 a_12
  4 b_13
  4 tpl_B1
  4 tpl_B2
  1 ds_db1Z
2 SimplifierDone 2


Linking M ...
-}


For anyone interested, the dependencies are twitter-conduit, conduit, lens.


--
Mateusz K.
_______________________________________________
Haskell-Cafe mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/haskell-cafe
Reply | Threaded
Open this post in threaded view
|

Re: GHC panic!

Ben Gamari-2
In reply to this post by 北原真一
北原真一 <[hidden email]> writes:

> I am asking for help I am Haskeller of Japan I'm using GHC7.6.3 GHC has
> caused the panic
> https://gist.github.com/minamiyama1994/6697c3fb9c345da6715f

I can reproduce the issue on my end with 7.6.3.  Setting
-fsimpl-tick-factor=10000 allows the build to proceed. This is a bit odd
as I don't see anything in your code that should give the simplifier too
much grief, but I could be missing something.  Unless you intend on
releasing the code I would probably add a 'Ghc-Options:
-fsimpl-tick-factor=1000' to your Cabal file for now.

Cheers,

- Ben

_______________________________________________
Haskell-Cafe mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/haskell-cafe

attachment0 (482 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: GHC panic!

Ben Gamari-2

CCing Michael in case this is something Conduit related that he
recognizes.

Ben Gamari <[hidden email]> writes:

> 北原真一 <[hidden email]> writes:
>
>> I am asking for help I am Haskeller of Japan I'm using GHC7.6.3 GHC has
>> caused the panic
>> https://gist.github.com/minamiyama1994/6697c3fb9c345da6715f
>
> I can reproduce the issue on my end with 7.6.3.  Setting
> -fsimpl-tick-factor=10000 allows the build to proceed. This is a bit odd
> as I don't see anything in your code that should give the simplifier too
> much grief, but I could be missing something.  Unless you intend on
> releasing the code I would probably add a 'Ghc-Options:
> -fsimpl-tick-factor=1000' to your Cabal file for now.
>
That being said, the Core here does look a bit concerning with blocks of
this nature being replicated 255 times (with both tick factors of 1000
and 2000),

    a2_rzJz
    a2_rzJz =
      \ @ a_X7Iy eta_X7IA _ _ eta3_B1 -> (# eta3_B1, eta_X7IA #)
   
    lvl24_rzJA
    lvl24_rzJA =
      IOError (Nothing) NoSuchThing getEnv3 getEnv2 (Nothing) lvl18_rzJs
   
    lvl25_rzJB
    lvl25_rzJB =
      \ @ a_a90p @ b_a90q m_a90r k_a90s r_X93X ->
        let {
          ds_X956
          ds_X956 = (m_a90r `cast` ...) r_X93X } in
        (\ r1_X95b s_X8Wq ->
           case (((ds_X956 `cast` ...) r1_X95b) `cast` ...) s_X8Wq
           of _ { (# ipv_X8SU, ipv1_X8SW #) ->
           ((((((k_a90s ipv1_X8SW) `cast` ...) r_X93X) `cast` ...) r1_X95b)
            `cast` ...)
             ipv_X8SU
           })
        `cast` ...

With associated case matches in `main`

I suppose it's posssible that this is intended, although the code does
seem quite bloated for such a simple program. Full Core here[1].

Cheers,

- Ben


[1] http://mw0.mooo.com/~ben/Main.core

_______________________________________________
Haskell-Cafe mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/haskell-cafe

attachment0 (482 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: GHC panic!

北原真一
In reply to this post by Ben Gamari-2
Thank you.
I do `-fsimpl-tick-factor=1024` and it is ok.
Thank you very much everyone.


2014-08-16 22:52 GMT+09:00 Ben Gamari <[hidden email]>:
北原真一 <[hidden email]> writes:

> I am asking for help I am Haskeller of Japan I'm using GHC7.6.3 GHC has
> caused the panic
> https://gist.github.com/minamiyama1994/6697c3fb9c345da6715f

I can reproduce the issue on my end with 7.6.3.  Setting
-fsimpl-tick-factor=10000 allows the build to proceed. This is a bit odd
as I don't see anything in your code that should give the simplifier too
much grief, but I could be missing something.  Unless you intend on
releasing the code I would probably add a 'Ghc-Options:
-fsimpl-tick-factor=1000' to your Cabal file for now.

Cheers,

- Ben


_______________________________________________
Haskell-Cafe mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/haskell-cafe