Re: Bug#351693: guile-1.6: builds on ia64, but causes other packages to FTBFS?

Re: Bug#351693: guile-1.6: builds on ia64, but causes other packages to FTBFS?

[Rob Browning]

Steve Langasek <vorlon@debian.org> writes:

> I see that guile-1.6 1.6.8-2 has now built successfully on ia64 (congrats!),
> but it looks like the first package to try to build *against* it, g-wrap,
> now fails with an error that looks like it's also a guile-1.6 problem:

[...]

> Rob, any ideas on this?

Indeed.  I can reproduce this in the sid chroot on merulo and after
some hacking, I've narrowed it down to a problem with Guile's
call-with-current-continuation (which is called by guile-library's
unit-test.scm[1]):

  (define (foo)
    (call-with-current-continuation
     (lambda (return)
       (return #t))))

  guile> (foo)
  Illegal instruction

I'm cc'ing upstream.


[1] Although Guile certainly needs to fix this bug, it looks like it
    might be easy to rewrite the unit-test.scm methods without using
    call-with-current-continuation, and at least with Guile, that
    should make them notably more efficient.

Thanks
-- 
Rob Browning
rlb @defaultvalue.org and @debian.org; previously @cs.utexas.edu
GPG starting 2002-11-03 = 14DD 432F AE39 534D B592  F9A0 25C8 D377 8C7E 73A4


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: Bug#351693: guile-1.6: builds on ia64, but causes other packages to FTBFS?

[Neil Jerram]

Rob Browning <rlb@defaultvalue.org> writes:

> Indeed.  I can reproduce this in the sid chroot on merulo and after
> some hacking, I've narrowed it down to a problem with Guile's
> call-with-current-continuation (which is called by guile-library's
> unit-test.scm[1]):
>
>   (define (foo)
>     (call-with-current-continuation
>      (lambda (return)
>        (return #t))))
>
>   guile> (foo)
>   Illegal instruction
>
> I'm cc'ing upstream.

Is anyone already working on this?  If not, I'll take a look.

Regards,
     Neil



_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: Bug#351693: guile-1.6: builds on ia64, but causes other packages to FTBFS?

[Rob Browning]

Neil Jerram <neil@ossau.uklinux.net> writes:

> Is anyone already working on this?  If not, I'll take a look.

I looked around a bit last night and found some relevant discussions:

  http://gcc.gnu.org/ml/gcc/2002-07/msg01417.html
  http://www.gelato.unsw.edu.au/archives/linux-ia64/0109/2178.html

I also came across this, which suggests that adding a dummy setjmp
before the getcontext call might help:

  http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/144939

So I tried adding a dummy setjmp call (which isn't actually executed)
right before getcontext in continuations.c, and that does fix the
illegal instruction when executing (call/cc (lambda (c) (c #t))).

However, without having a better understanding about what's actually
happening, I'm not inclined to treat this as a correct solution yet.

Can anyone else comment more conclusively?

Thanks
-- 
Rob Browning
rlb @defaultvalue.org and @debian.org; previously @cs.utexas.edu
GPG starting 2002-11-03 = 14DD 432F AE39 534D B592  F9A0 25C8 D377 8C7E 73A4


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: Bug#351693: guile-1.6: builds on ia64, but causes other packages to FTBFS?

[Rob Browning]

Rob Browning <rlb@defaultvalue.org> writes:

> I also came across this, which suggests that adding a dummy setjmp
> before the getcontext call might help:
>
>   http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/144939

Here's a more recent comment which appears to say that the dummy
setjmp should actually be after the call to getcontext:

  http://lists.freebsd.org/pipermail/freebsd-sparc64/2006-January/003739.html

-- 
Rob Browning
rlb @defaultvalue.org and @debian.org; previously @cs.utexas.edu
GPG starting 2002-11-03 = 14DD 432F AE39 534D B592  F9A0 25C8 D377 8C7E 73A4


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel

Re: Bug#351693: guile-1.6: builds on ia64, but causes other packages to FTBFS?

[Rob Browning]

I've committed the fix suggested in the following thread to 1.6, 1.8,
and CVS head:

  http://lists.gnu.org/archive/html/guile-devel/2006-07/msg00006.html

I'll try to release updated Debian packages soon.

Thanks
-- 
Rob Browning
rlb @defaultvalue.org and @debian.org; previously @cs.utexas.edu
GPG starting 2002-11-03 = 14DD 432F AE39 534D B592  F9A0 25C8 D377 8C7E 73A4


_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-devel