segfault in compute_transp()

[Daniel Bayer]

Moin,

On Tue, Sep 21, 2004 at 06:04:14PM +0200, Steven Bosscher wrote:
> On Tuesday 21 September 2004 17:33, Daniel Bayer wrote:
> > I'm writing a new backend for the gcc. (...)
> > This insn was generated in the addressof-pass.
> 
> Then you are developing against an old tree.  There is no
> addressof pass in mainline anymore.  It was removed months
> ago.

At least 3.4.2 still has this pass and the problem as described in my
last mail. But as I have some more problems that seem to be related to
this pass I will try to update.

> You might want to consider working against a more recent
> snapshot.

I tried the latest snapshot but had some problems[0] bootstrapping it
on both, my pc and my u10. Tomorrow I will have a more closer look on
it.

But my question still remains. Are INSNs without a blocknumber legal
output of a pass or not?


Daniel

[0] Make had some problems finding install-sh and other files.