Eclipse-epic build fails on ppc and ppc64 for F8 and F10

[Mat Booth]

Updating my packages today I'm finding that Java is crashing when
building eclipse-epic on ppc/ppc64 on F8 and F10 with:

*** glibc detected *** java: free(): invalid next size (normal): 0x10113540 ***

F10 ppc stack trace:
http://koji.fedoraproject.org/koji/getfile?taskID=661999&name=build.log

F8 ppc64 stack trace:
http://koji.fedoraproject.org/koji/getfile?taskID=661990&name=build.log

Strangely, it worked fine on F9.

Is this a known bug?

-- 
Mat Booth
www.matbooth.co.uk

[Andrew Haley]

Mat Booth wrote:
>  Updating my packages today I'm finding that Java is crashing
> when
>  building eclipse-epic on ppc/ppc64 on F8 and F10 with:
>  
>  *** glibc detected *** java: free(): invalid next size (normal): 0x10113540 ***
>  
>  F10 ppc stack trace:
>  http://koji.fedoraproject.org/koji/getfile?taskID=661999&name=build.log
>  
>  F8 ppc64 stack trace:
>  http://koji.fedoraproject.org/koji/getfile?taskID=661990&name=build.log
>  
>  Strangely, it worked fine on F9.
>  
>  Is this a known bug? 
Certainly not.  It think must have built on f8.

If this is repeatable then I'll have to debug it.  I don't have easy access
to a ppc machine, unfortunately, but perhaps we can work something out.

It's failing inside dlopen(), which is weird.  Do you know if gdb on ppc
has watchpoints?  Or Valgrind?

Andrew.

[Mat Booth]

On Sat, Jun 14, 2008 at 8:48 PM, Andrew Haley <aph(a)redhat.com&gt; wrote:
>  Mat Booth wrote:
> > Updating my packages today I'm finding that Java is crashing when
> > building eclipse-epic on ppc/ppc64 on F8 and F10 with:
> >
> > *** glibc detected *** java: free(): invalid next size (normal): 0x10113540 ***
> >
> > F10 ppc stack trace:
> > http://koji.fedoraproject.org/koji/getfile?taskID=661999&name=build.log
> >
> > F8 ppc64 stack trace:
> > http://koji.fedoraproject.org/koji/getfile?taskID=661990&name=build.log
> >
> > Strangely, it worked fine on F9.
> >
> > Is this a known bug?
>
>  Certainly not.  It think must have built on f8.
>
>  If this is repeatable then I'll have to debug it.  I don't have easy access
>  to a ppc machine, unfortunately, but perhaps we can work something out.
>
>  It's failing inside dlopen(), which is weird.  Do you know if gdb on ppc
>  has watchpoints?  Or Valgrind?
>
>  Andrew.
>
>
>  
Weird, I tried it again this morning and it succeeded on F8:
http://koji.fedoraproject.org/koji/taskinfo?taskID=662565

I'm afraid I don't have access to a ppc machine at all, so I can't
answer your other questions.

-- 
Mat Booth
www.matbooth.co.uk

[Gary Benson]

Andrew Haley wrote:
>  If this is repeatable then I'll have to debug it.  I don't
> have easy
>  access to a ppc machine, unfortunately, but perhaps we can work
>  something out. 
There is tofu in Toronto, or I can probably make you an account on the
machine I use.  I have no idea who owns that box though now that David
Woodhouse left...

Cheers,
Gary

--
http://gbenson.net/