All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jim Fehlig <jfehlig@novell.com>
To: Keir Fraser <keir@xensource.com>
Cc: xen-devel@lists.xensource.com
Subject: Re: [PATCH] [xm] Fix vncdisplay for hvm guests
Date: Thu, 31 May 2007 16:26:38 -0600	[thread overview]
Message-ID: <465F4B9E.7020302@novell.com> (raw)

Keir Fraser wrote:
>
> On 16/5/07 00:14, "Jim Fehlig" <jfehlig@novell.com> wrote:
>
>   
>> results in '-vncunused' being passed to qemu-dm.  There are several
>> approaches
>> for a fix - this patch defaults vncdisplay to None in xm options.  It
>> currently defaults to 1 and is always included in the image config
>> created by configure_hvm() in tools/python/xen/xm/create.py.  In xend
>> (tools/python/xen/xend/image.py - parseDeviceModelArgs), vncunused takes
>> precedence over vncdisplay.
>>     
>
> Looks like it changes vncunused default rather than vncdisplay. Wouldn't the
> preferred default be to keep vncunused=1?
>   

After looking at this closer I thing the original patch applies. 
Setting a default in xm tool doesn't seem right.  E.g. with hvm config

vnc=1
vncdisplay=5

I get two different versions of xend's internal config for vfb device -
depending on client I use to define the domain.  Using 'xm new config', 
/var/lib/xend/domains/<uuid>/config.sxp has

        (vfb
            (vncunused 1)
            (vncdisplay 5)
            (uuid 499604ae-f8c5-81a6-3600-9444322e2bfc)
        )

Using XenAPI c-bindings I get

       (vfb
            (type vnc)
            (protocol rfb)
            (uuid 66c18754-3207-5e45-31db-28df050bff4f)
            (vndisplay 5)
       )

So if we want a default it should be in xend for consistency.

Now as for default I found some interesting behavior using vncdisplay on
c/s 15080.  For hvm domains created with xm (containing above config),
the following qemu-dm cmdline is assembled

-vnc 127.0.0.1:5 -vncunused

If another domain is using 5905 the new domain will bind to 5906 due to
the -vncunsued also being present on cmdline, otherwise it will bind to
5905 as expected.

The story is a little different for pv domains.  A pv domain 'xm new'ed'
with config

vfb=["type=vnc,vncdisplay=5"]

results in /var/lib/xend/domains/<uuid>/config.sxp

        (vfb
            (xauthority /root/.Xauthority)
            (vncdisplay 5)
            (type vnc)
            (display localhost:10.0)
            (uuid 5741017b-f0fc-0447-c613-aa558f6e582c)
        )

Notice there is no vncunused=1 in this config as there was in the
internal hvm config.  xen-vncfb is invoked with "--vncport 5905 --listen
127.0.0.1".  If another domain is already using 5905, too bad -
xen-vncfb won't be able to bind and no graphics.

Which of these behaviors is preferred default?  I can put together a
patch that provides consistency between hvm and pv domains once default
is chosen.  Personally I'm torn.  If user specifies a port she should be
able to reach the display at that port.  On the other hand, having a
functional vm in the event of a conflict is nice too :-).
 
Regards,
Jim

             reply	other threads:[~2007-05-31 22:26 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-05-31 22:26 Jim Fehlig [this message]
2007-06-25 13:53 ` [PATCH] [xm] Fix vncdisplay for hvm guests jd
  -- strict thread matches above, loose matches on Subject: below --
2007-05-24 16:42 Jim Fehlig
2007-05-15 23:14 Jim Fehlig
2007-05-24 14:31 ` Keir Fraser

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=465F4B9E.7020302@novell.com \
    --to=jfehlig@novell.com \
    --cc=keir@xensource.com \
    --cc=xen-devel@lists.xensource.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.