xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
From: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
To: "Pavel Matěja" <pavel@netsafe.cz>
Cc: xen-devel@lists.xen.org
Subject: Re: ATI/AMD VGA passthru report
Date: Wed, 9 May 2012 13:44:26 -0400	[thread overview]
Message-ID: <20120509174426.GD9094@phenom.dumpdata.com> (raw)
In-Reply-To: <201205081312.39927.pavel@netsafe.cz>

On Tue, May 08, 2012 at 01:12:39PM +0200, Pavel Matěja wrote:
> On Tue 8. of May 2012 01:07:07 Konrad Rzeszutek Wilk wrote:
> > On Tue, May 08, 2012 at 01:10:32AM +0200, Pavel Mateja wrote:
> > > > On Mon, May 07, 2012 at 01:45:23PM +0200, Pavel Matěja wrote:
> > > >> Hi,
> > > >> I'm trying to run AMD VGA passthru on latest XEN unstable with latest
> > > >> kernel.
> > > >> I already have working setup with ancient xen-devel 4.2 and 2.6.32.x
> > > >> xenified
> > > >> kernel. See attached log.
> > > >> 
> > > >> So I tried just to update and patch xen and kernel, but I had no luck
> > > >> so far.
> > > >> 
> > > >> Xen has ati_vbios_patch_respin.txt sent to xen-devel by Wei Huang
> > > >> recently.
> > > >> http://lists.xen.org/archives/html/xen-devel/2012-04/msg00291.html
> > > >> 
> > > >> I tried two kernels
> > > >> testing-3.5-with-extra and xen/next-3.2 from
> > > >> git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git
> > > >> and
> > > >> git://git.kernel.org/pub/scm/linux/kernel/git/jeremy/xen.git
> > > >> Both with ioperm opcode patch which is required by the ATI patch. See
> > > >> attachement.
> > > >> 
> > > >> The xen/next-3.2 branch kernel was able to start booting windows.
> > > >> With Catalyst driver installed I just saw bluescreen during Windows
> > > >> boot.
> > > >> Without Catalyst driver Windows were able to boot but Windows ended
> > > >> frozen or
> > > >> USB was not working. It was hard to tell because input devices had no
> > > >> response
> > > >> at all.
> > > >> 
> > > >> The testing-3.5-with-extra (reported as 3.4.0-rc4) just crashed dom0
> > > >> kernel
> > > >> during Windows boot. I guess I have to rework the io_bitmap patch
> > > >> somehow.
> > > > 
> > > > When you say 'io_bitmap' patch are you referring to these ones:
> > > > 
> > > > 70a357d xen: implement IO permission bitmap
> > > > 0c596c5 x86/paravirt: paravirtualize IO permission bitmap
> > > > 93b7a2a x86: demacro set_iopl_mask()
> > > > 
> > > > ? I don't think those are in that tag. They are in the kitchensink -
> > > > #testing
> > > > - does it work with that branch?
> > > > 
> > > > The next-3.2 is a bit old. I should actually delete it.
> > > 
> > > Hi,
> > > I mean this patch below.
> > > I think it was originally here:
> > > http://old-list-archives.xen.org/archives/html/xen-devel/2009-05/msg01139
> > > .html Then I was told by you to use devel/ioperm branch which I think
> > > doesn't exist anymore:
> > > http://old-list-archives.xen.org/archives/html/xen-devel/2011-11/msg01213
> > > .html
> > 
> > It certainly does :-)
> > 
> > > I'm too tired to test #testing branch now. I'll try it tomorrow after
> > > some sleep.
> > 
> > OK, take your time.
> 
> Hi again!
> I did first tests with xen-unstable and the #testing kernel.
> 
> Windows showed blue screen during boot with xm toolstack.
> 
> But I had luck with xl. Windows booted succesfully so I ran the experience 
> index tests:

> kernel:	2.6.32.57	#testing
> CPU		7,4		7,4
> RAM		7,4		5,5
> Aero		7,1		7,1
> Gr.card	7,1		7,1
> Disk		7,3		7,5

So the issue was that 'xm' didn't work but 'xl' did?

> I tried to run some games as well but the feeling is much worse compared to 
> older configuration. Like lags..

Do the lags disappear if you pin the vCPUs?
> 
> Are there some older branches which are supposed to work?

There is not a specific branch per say. There are some auxillary branches
with patches that enable this and that. So what might work is
to use my #linux-next branch and merge both devel/misc and devel/ioperm
in and see what happens. If I get some time today I will try that and
stick it as tag.

> 
> I tried xen-4.1.3-rc1 with #testing kernel but it just freezes after while 
> even with Dom0 running. Only sys-rq worked.

Oh no!
> 
> I have second slightly newer system too so I'll retest that one later.
> -- 
> Pavel Mateja
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

  reply	other threads:[~2012-05-09 17:44 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-05-07 11:45 ATI/AMD VGA passthru report Pavel Matěja
2012-05-07 13:46 ` Pavel Matěja
2012-05-07 17:31 ` Konrad Rzeszutek Wilk
2012-05-07 23:10   ` Pavel Mateja
2012-05-07 23:07     ` Konrad Rzeszutek Wilk
2012-05-08 11:12       ` Pavel Matěja
2012-05-09 17:44         ` Konrad Rzeszutek Wilk [this message]
2012-05-09 18:25           ` Pavel Matěja
2012-05-09 18:59           ` Pavel Matěja
2012-05-09 19:18             ` djmagee

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=20120509174426.GD9094@phenom.dumpdata.com \
    --to=konrad.wilk@oracle.com \
    --cc=pavel@netsafe.cz \
    --cc=xen-devel@lists.xen.org \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).