[coyotos-dev] [The EROS Group] Request
Jonathan S. Shapiro
shap at eros-os.com
Mon Oct 1 14:55:18 EDT 2007
On Mon, 2007-10-01 at 19:06 +0200, Valerio Bellizzomi wrote:
> On 01/10/2007, at 9.52, Jonathan S. Shapiro wrote:
>
> >On Mon, 2007-10-01 at 05:47 -0500, devbox at selnet.org wrote:
> >
> >> While you are working at the SMP initialization code, can you put a
> >> message in the console demo to display the type and number of CPUs (so
> >> that we can see if all CPUs are being recognized) ?
> >
> >We can't do those messages until we actually IPL those CPUs. Right now
> >that isn't high priority.
> >
> >> Second thing: I know how to set up serial console in GRUB, however it
> >> would currently require that I re-apply the patch every time I rebuild.
> >It
> >> would be nice if you apply this patch in the tree.
> >
> >So just to be clear, the patch you want is to the *grub* config? You
> >understand that the kernel will still be talking to the frame buffer
> >unless HAVE_HUI is disabled, right? And if HAVE_HUI is disabled you
> >won't get *any* output.
>
> Can this logic be revised?
>
> I really would like to avoid copying console output by hand when I report
> tests, this means you will continue to receive photos of the console.
> Well, I can live with photos at the moment.
> In any case I think the option to redirect console output to a serial
> terminal is definitely a must...
So first, I disagree on the "must". Most early platforms will have no
console output at all.
Regarding debug capture, are you aware that there is already a GDB
script procedure for dumping the ring buffer? Take a look at
print-message-buffer in the arch/i386/scripts/coyotos.gdb file...
--
Jonathan S. Shapiro
Managing Director
The EROS Group, LLC
www.coyotos.org, www.eros-os.org
More information about the coyotos-dev
mailing list