Re: Regression: patch " hvc_console: display printk messages on console." causing infinite loop with 3.2-rc0 + Xen.

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Thu, Oct 27, 2011 at 07:45:07AM -0700, Miche Baker-Harvey wrote:
> Bummer.  Sorry about that.
> 
> Konrad, could you give me a hand figuring out what's up with this - I'm not
> running Zen here.  (If there's a test suite I could have run, mea culpa, and
> please point me in the right direction.)

I kind of run the nightly build/test rinse & repeat cycle using this:

git://xenbits.xen.org/xentesttools/bootstrap.git

But it might be just easier for you to install Fedora Core 16, do
'yum install xen', reboot, make sure you can boot Xen, and then build
a 3.2 kernel with your patch + debug stuff.

> 
> The "console [hvc0] enabled" message is coming from register_console() in
> printk.c, and it doesn't have any loops.
> It's called from hvc_console_init(), hvc_instantiate(), and hvc_alloc(),
> none of which have loops.  The call in hvc_alloc() is the one I added, so
> either I've inadvertently made one of the other two repeat (I'm guestting
> hvc_instantiate() is the one you're using), or you're infinitely calling
> hvc_alloc().  There's a calls to hvc_alloc() in hvc_xen.c::xen_hvc_init(),
> and in init_port_console() in virtio_console.c, but again, no loops.
>  Similarly for hvc_instantiate().
> 
> So, could I trouble you to put a breakpoint in register_console() and send
> me the stack?

A dump_stack() should suffice I presume? Yeah let me do that.