Re: fix for bug 2969

[Rick Moseley <rmoseley at redhat dot com>]

Yao Qi wrote:
> On Mon, Jul 31, 2006 at 01:43:41PM +0800, Yao Qi wrote:
>   
> > libunwind is not supported on ppc/ppc64, so libuwind and its java
> > bindings could be skipped when build frysk on ppc/ppc64.
> >
> > 2006-07-31  Yao Qi  <qiyaoltc@cn.ibm.com>
> >
> > 	* frysk-imports/bootstrap.sh: Build libunwind if it is supported.
> >
> > Run bootstrap.sh on X86 and PPC64 to verify its output(Makefile.gen),
> > and libunwind and its java bindings are only built on X86.
> >
> > OK to apply?  Thanks.
> >
> >     
>  
> Files in frysk-core/frysk/rt/ also do some java bindings for libunwind,
> and this patch could not cover it.
>
> Could we put these libunwind java bindings together in the same place?
> such as frysk-import/lib/unwind.  That would be easy to organize and 
> maintain the code about libunwind.  Code in frysk-core/frysk/rt only
> need to call the methods provided by libunwind java-bindings.
>
> Any comments?  Thanks!
>   
Yao,

I saw the exchange on IRC last evening between Wu and Andrew and it 
appears they have come to a resolution on these issues.  For the rest of 
the world and myself, here is my understanding of the resolutions:

files in frysk-core/frysk/rt cause the ppc/ppc64 build to fail:

   Resolution:  IBM will write stubs for libunwind to fake ppc/ppc64 
capabilities until such time as those arches are supported leaving the 
code where it is for right now

libiberty/libbfd issue:

   Resolution:  Red Hat will provide new "cleanroom" code for this to 
be incorporated into Frysk next week


Please correct me if I have any of this wrong.

Rick