Re: hppa and haskell packages
On Wed, Mar 24, 2010 at 05:30:43PM +0100, Joachim Breitner wrote:
> Hi,
>
> I???m well aware that some haskell packages are quite heavy to build. But
> I???m surprised to see hppa fail where arches like armel and mips* cope.
> Is there something else broken?
>
> These packages??? build failure on hppa prevent a testing migration:
>
> hlint (hppa, needs haskell-src-exts)
>
> These packages could be added to testing if they would build on hppa:
>
> haskell-regex-tdfa (hppa, mipsel)
> haskell-src-exts (hppa, armel, mips, mipsel, s390)
>
> These packages never built on hppa, so the build failure does not hinder
> the migrations:
>
> xmonad-contrib
> agda
>
> Can you possibly try the builds again on a stronger hppa buildd box, if
> available?
I started an agda build on paer 8 days ago - it is still running. Memory
does not appear to be an issue (swap is available, but unused). Throwing
more CPU at it might improve things - but I don't think a CPU exists with
enough cycles to make it build in any reasonable amount of time.
strace of the ghc6 process shows an infinite loop of:
rt_sigreturn() = -14059
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x40002b28) = -513
--- SIGVTALRM (Virtual timer expired) @ 0 (0) ---
rt_sigreturn() = -14059
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x40002b28) = -513
--- SIGVTALRM (Virtual timer expired) @ 0 (0) ---
rt_sigreturn() = -14059
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x40002b28) = -513
--- SIGVTALRM (Virtual timer expired) @ 0 (0) ---
rt_sigreturn() = -14059
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x40002b28) = -513
--- SIGVTALRM (Virtual timer expired) @ 0 (0) ---
rt_sigreturn() = -14059
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x40002b28) = -513
--- SIGVTALRM (Virtual timer expired) @ 0 (0) ---
rt_sigreturn() = -14059
I don't think the hardware is the issue here - this has got to be a
bug somewhere.
--
dann frazier
Reply to: