Re: [Xen-ia64-devel] serial console settings on Tiger4

[Horms <horms@xxxxxxxxxxxx>]

On Wed, Feb 14, 2007 at 08:33:57AM -0700, Alex Williamson wrote:
> Hi Simon,
> 
> On Wed, 2007-02-14 at 18:42 +0900, Horms wrote:
> > 
> > Being greeted by a lovely not-working console this afternoon
> 
>    Sorry about that, thanks for testing.
> 
> > I have set about testing this on Tiger2. Here are the results:
> > 
> > Current tree (8bdbe88e422f):          Broken
> > Current tree - 7efb3a06c56a:          Ok
> > Current tree + Alex's first patch:    Ok
> > Current tree + Alex's first patch
> >              + Alex's second patch:   Ok
> > Current tree + Alex's first patch
> >              + Alex's second patch
> >          + Kawamura-san's patch:  Ok
> > 
> > So in short, while the current tree is broken, it
> > seems pretty easy to make Tiger2 work.
> > 
> > All of these tests were made with the following settings
> > (ignore the irrelevant ones :)
> > 
> > append="maxcpus=1 phys_efi com2=115200,8n1,0x2f8,0 console=com2
>                              ^^^^^^^^^^^^^^^^^^^^^^^
> 
> This should be unnecessary with all the patches applied.  For a Tiger2,
> com1 and com2 should be installed as 0x3f8 and 0x2f8 respectively.  The
> ns16550 driver _should_ be able to read the baud rate out of the UART.
> 
> >         conswitch=bb sync_console dom0_max_vcpus=2 dom0_mem=1G 
> > crashkernel=256M
> >         -- clock=pit ip=on apm=power-off console=tty0 console=ttyS0,115200"
>                                                       unnecessary -> ^^^^^^^
> 
> It's interesting that even on the Tiger2 you're using 0x2f8 as your
> console.  Is that typical?  I'm wondering if we should handle both
> Tiger2 and Tiger4 the same way and only setup com1 as 0x2f8.  Then you
> wouldn't even need the console=com2 option.  Is 0x3f8 usable as a
> console on either box?  My impression from Kawamura-san was that Tiger4
> didn't have a physical connection for 0x3f8, but Tiger2 might.  Thanks,

Hi Alex,

I tried removing the "^^^^^^" portions above as you suggested,
and yes, everything still works fine. I believe that my Tiger2 setup
was derived from a Tiger4 setup (I swapped machines with Yamahata-san
at some stage), so that probably explains most of the weirdness.

As for accessing 0x3f8 (com2?). I'm not really sure. It may well exists,
but last time I inspected the box it seemed to only have one console
output - though it is through an R45 IIRC, so perhaps that exposes
two ports. I can investigate further.

And as for getting rid of console=com2 by wiring com1 as 0x2f8,
that would probably work quite well for me. Though I guess that
if 0x3f8 actually exists (and is usable somewhow?) then its probably not
the way to go.

I see that you checked in the fix overnight. I'll also double-check
that it still works on Tiger2.

-- 
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/


_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel