More on the armv4l segfault.

[Rob Landley]

On Thursday 14 December 2006 6:50 pm, Erik Andersen wrote:
> On Thu Dec 14, 2006 at 06:45:49PM -0500, Rob Landley wrote:
> > In order to go from 16820 to 16827 and get something that compiles, I have 
to 
> > apply 16821, 16823, 16825, 16826, and 16827.  And that produces the 
segfault 
> > when I run a "hello world" built from the result.
> > 
> > This at least means it's not 16822 or 16824.  All five of the above 
patches 
> > are from Erik, so it's something he did with the thread cleanups...
> 
> Does it still blow up when you link the hello world app with
> libpthread?

Makes no difference.

I think I've narrowed it down to the 16825/16827 pair.  If I add 16821 and 
16826, it works.  If I add 16823 to that, it still works.  If I add 16825 to 
that, I get a toolchain that fails to link.  (Whether or not I 
add -lpthread.)  If I add 16827, I get the segfault on exit.

So 16825/16827 is breaking armv4l.

Any ideas?

Rob
-- 
"Perfection is reached, not when there is no longer anything to add, but
when there is no longer anything to take away." - Antoine de Saint-Exupery