[Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[Juergen Lock]

Hmm.  Looks like the latest vmwarevga commits broke *ix/xorg guests.
I tested the 2008-03-11 cvs snapshot with
sidux-2008-01-200803010113-nyx_pre1-kde-lite-i386.iso (debian sid based
Linux livecd) and FreeSBIE-2.0.1-RELEASE.iso (FreeBSD based livecd),
on sidux xorg said:

(WW) ****INVALID MEM ALLOCATION**** b: 0xf0000000 e: 0xf07fffff correcting

 and the xserver display was garbage, and on FreeSBIE xorg didn't even
start, showing a similar complaint. (both with Driver "vmware" in xorg.conf)
Do the latest commits need a bios update maybe?

 Anyway, a 2008-03-02 snapshot works (now committed as FreeBSD qemu-devel
port), at least for these two guests (I don't have a windows guest here atm.)
FreeSBIE even was able to attach ed0 (the default ne2kpci nic) as well as
the emulated es1370 soundcard, which didn't work previously when using
vmwarevga.

 Just thought I'd mention... :)
	Juergen

Re: [Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[Rick Vernam]

On Wednesday 12 March 2008 04:10:24 pm Juergen Lock wrote:
> Hmm.  Looks like the latest vmwarevga commits broke *ix/xorg guests.
> I tested the 2008-03-11 cvs snapshot with
> sidux-2008-01-200803010113-nyx_pre1-kde-lite-i386.iso (debian sid based
> Linux livecd) and FreeSBIE-2.0.1-RELEASE.iso (FreeBSD based livecd),
> on sidux xorg said:
>
> (WW) ****INVALID MEM ALLOCATION**** b: 0xf0000000 e: 0xf07fffff correcting
...
>  Just thought I'd mention... :)
> 	Juergen

any word on this?
I can also report the same result on W2K & WXP guests (both using drivers from 
most recent vmware-workstation release).
in windows' device manager there is a similar message about that memory 
address.

I reverted this patch, which restored a working vmware vga device
http://cvs.savannah.gnu.org/viewcvs/qemu/hw/vmware_vga.c?cvsroot=qemu&r1=1.7&r2=1.8
however, with this patch reverted I get occasional triple-faults.

Regretably, with the time available to me, I've not yet been able to get a 
deep enough grasp to do any useful debugging...but I'd be happy to test 
anything out...

Re: [Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[WaxDragon]

On Thu, Apr 3, 2008 at 8:12 PM, Rick Vernam <rickv@hobi.com> wrote:
> On Wednesday 12 March 2008 04:10:24 pm Juergen Lock wrote:
>  > Hmm.  Looks like the latest vmwarevga commits broke *ix/xorg guests.
>  > I tested the 2008-03-11 cvs snapshot with
>  > sidux-2008-01-200803010113-nyx_pre1-kde-lite-i386.iso (debian sid based
>  > Linux livecd) and FreeSBIE-2.0.1-RELEASE.iso (FreeBSD based livecd),
>  > on sidux xorg said:
>  >
>  > (WW) ****INVALID MEM ALLOCATION**** b: 0xf0000000 e: 0xf07fffff correcting
>  ...
>
> >  Just thought I'd mention... :)
>  >       Juergen
>
>  any word on this?
>  I can also report the same result on W2K & WXP guests (both using drivers from
>  most recent vmware-workstation release).
>  in windows' device manager there is a similar message about that memory
>  address.
>
>  I reverted this patch, which restored a working vmware vga device
>  http://cvs.savannah.gnu.org/viewcvs/qemu/hw/vmware_vga.c?cvsroot=qemu&r1=1.7&r2=1.8
>  however, with this patch reverted I get occasional triple-faults.

Confirmed, reverting hw/vmware_vga.c to r4022 restores x.org and
framebuffer functionality.

>
>  Regretably, with the time available to me, I've not yet been able to get a
>  deep enough grasp to do any useful debugging...but I'd be happy to test
>  anything out...

4023's x.org output:
http://waxdragon.homeip.net/~ford/qemu.Xorg.0.log

4022's lspci output:

00:02.0 VGA compatible controller: VMware Inc [VMware SVGA II] PCI
Display Adapter (prog-if 00 [VGA])
        Subsystem: VMware Inc [VMware SVGA II] PCI Display Adapter
        Flags: bus master, fast devsel, latency 64
        I/O ports at c010 [size=16]
        Memory at 20000000 (32-bit, non-prefetchable) [size=512M]

4023's lspci output:

00:02.0 VGA compatible controller: VMware Inc [VMware SVGA II] PCI
Display Adapter (prog-if 00 [VGA])
    Subsystem: VMware Inc [VMware SVGA II] PCI Display Adapter
    Flags: bus master, fast devsel, latency 64
    Memory at f0000000 (32-bit, prefetchable) [size=8M]
    Memory at 20000000 (32-bit, non-prefetchable) [size=512M]

and for reference, vmware workstation 6's lspci:

0000:00:0f.0 VGA compatible controller: VMWare Inc [VMWare SVGA II]
PCI Display Adapter (prog-if 00 [VGA])
    Subsystem: VMWare Inc [VMWare SVGA II] PCI Display Adapter
    Flags: medium devsel
    I/O ports at 10a0 [size=16]
    Memory at f0000000 (32-bit, non-prefetchable) [size=128M]
    Memory at e8000000 (32-bit, non-prefetchable) [size=8M]
    Expansion ROM at <unassigned> [disabled] [size=32K]





-- 
18:22 <@GreatLord> dx8 or dx9 goes over opengl
18:22 <@GreatLord> with wine warper
18:22 <@silverblade> WARP FACTOR WINE
18:25 <@WaxDragon> Set the controls for the heart of the LULS.

Re: [Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[Aurelien Jarno]

On Fri, Apr 04, 2008 at 07:59:02AM -0400, WaxDragon wrote:
> On Thu, Apr 3, 2008 at 8:12 PM, Rick Vernam <rickv@hobi.com> wrote:
> > On Wednesday 12 March 2008 04:10:24 pm Juergen Lock wrote:
> >  > Hmm.  Looks like the latest vmwarevga commits broke *ix/xorg guests.
> >  > I tested the 2008-03-11 cvs snapshot with
> >  > sidux-2008-01-200803010113-nyx_pre1-kde-lite-i386.iso (debian sid based
> >  > Linux livecd) and FreeSBIE-2.0.1-RELEASE.iso (FreeBSD based livecd),
> >  > on sidux xorg said:
> >  >
> >  > (WW) ****INVALID MEM ALLOCATION**** b: 0xf0000000 e: 0xf07fffff correcting
> >  ...
> >
> > >  Just thought I'd mention... :)
> >  >       Juergen
> >
> >  any word on this?
> >  I can also report the same result on W2K & WXP guests (both using drivers from
> >  most recent vmware-workstation release).
> >  in windows' device manager there is a similar message about that memory
> >  address.
> >
> >  I reverted this patch, which restored a working vmware vga device
> >  http://cvs.savannah.gnu.org/viewcvs/qemu/hw/vmware_vga.c?cvsroot=qemu&r1=1.7&r2=1.8
> >  however, with this patch reverted I get occasional triple-faults.
> 
> Confirmed, reverting hw/vmware_vga.c to r4022 restores x.org and
> framebuffer functionality.
> 

I have just fixed an obvious bug. Please test if it fixes the problem.


-- 
  .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
 : :' :  Debian developer           | Electrical Engineer
 `. `'   aurel32@debian.org         | aurelien@aurel32.net
   `-    people.debian.org/~aurel32 | www.aurel32.net

Re: [Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[Rick Vernam]

On Friday 04 April 2008 12:17:34 pm Aurelien Jarno wrote:
> I have just fixed an obvious bug. Please test if it fixes the problem.

Not on w2k or wxp guests with latest vmware drivers.

I don't have any linux guests with X, so I'm grabbing a live-cd right now to 
try it out there.

Re: [Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[andrzej zaborowski]

Hi,

On 04/04/2008, Rick Vernam <rickv@hobi.com> wrote:
> On Wednesday 12 March 2008 04:10:24 pm Juergen Lock wrote:
>  > Hmm.  Looks like the latest vmwarevga commits broke *ix/xorg guests.
>  > I tested the 2008-03-11 cvs snapshot with
>  > sidux-2008-01-200803010113-nyx_pre1-kde-lite-i386.iso (debian sid based
>  > Linux livecd) and FreeSBIE-2.0.1-RELEASE.iso (FreeBSD based livecd),
>  > on sidux xorg said:
>  >
>  > (WW) ****INVALID MEM ALLOCATION**** b: 0xf0000000 e: 0xf07fffff correcting
>
> ...
>
> >  Just thought I'd mention... :)
>  >       Juergen
>
>
> any word on this?
>  I can also report the same result on W2K & WXP guests (both using drivers from
>  most recent vmware-workstation release).
>  in windows' device manager there is a similar message about that memory
>  address.
>
>  I reverted this patch, which restored a working vmware vga device
>  http://cvs.savannah.gnu.org/viewcvs/qemu/hw/vmware_vga.c?cvsroot=qemu&r1=1.7&r2=1.8
>  however, with this patch reverted I get occasional triple-faults.

Can you try with the above patch reverted and this one:
http://lists.gnu.org/archive/html/qemu-devel/2008-02/msg00415.html
applied in place of it?  The one I applied was supposed to be
equivalent but I likely screwed it up.

I would love to test it myself but I won't have an opportunity for the moment.
-- 
Please do not print this email unless absolutely necessary. Spread
environmental awareness.

Re: [Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[Rick Vernam]

On Friday 04 April 2008 01:44:42 pm andrzej zaborowski wrote:
> Hi,
>
> On 04/04/2008, Rick Vernam <rickv@hobi.com> wrote:
> > On Wednesday 12 March 2008 04:10:24 pm Juergen Lock wrote:
> >  > Hmm.  Looks like the latest vmwarevga commits broke *ix/xorg guests.
> >  > I tested the 2008-03-11 cvs snapshot with
> >  > sidux-2008-01-200803010113-nyx_pre1-kde-lite-i386.iso (debian sid
> >  > based Linux livecd) and FreeSBIE-2.0.1-RELEASE.iso (FreeBSD based
> >  > livecd), on sidux xorg said:
> >  >
> >  > (WW) ****INVALID MEM ALLOCATION**** b: 0xf0000000 e: 0xf07fffff
> >  > correcting
> >
> > ...
> >
> > >  Just thought I'd mention... :)
> > >
> >  >       Juergen
> >
> > any word on this?
> >  I can also report the same result on W2K & WXP guests (both using
> > drivers from most recent vmware-workstation release).
> >  in windows' device manager there is a similar message about that memory
> >  address.
> >
> >  I reverted this patch, which restored a working vmware vga device
> > 
> > http://cvs.savannah.gnu.org/viewcvs/qemu/hw/vmware_vga.c?cvsroot=qemu&r1=
> >1.7&r2=1.8 however, with this patch reverted I get occasional
> > triple-faults.
>
> Can you try with the above patch reverted and this one:
> http://lists.gnu.org/archive/html/qemu-devel/2008-02/msg00415.html
> applied in place of it?  The one I applied was supposed to be
> equivalent but I likely screwed it up.
>
> I would love to test it myself but I won't have an opportunity for the
> moment.

I'll try in a few hours.
Thanks much...

Re: [Qemu-devel] vmwarevga regression (broken with two Linux and FreeBSD guests)

[Rick Vernam]

On Friday 04 April 2008 01:44:42 pm andrzej zaborowski wrote:
[snip]
> >  > (WW) ****INVALID MEM ALLOCATION**** b: 0xf0000000 e: 0xf07fffff
> >  > correcting
[snip]
> >
> >  I reverted this patch, which restored a working vmware vga device
> > http://cvs.savannah.gnu.org/viewcvs/qemu/hw/vmware_vga.c?cvsroot=qemu&r1=
> >1.7&r2=1.8 however, with this patch reverted I get occasional
> > triple-faults.
>
> Can you try with the above patch reverted and this one:
> http://lists.gnu.org/archive/html/qemu-devel/2008-02/msg00415.html
> applied in place of it?  The one I applied was supposed to be
> equivalent but I likely screwed it up.
>

Win 2K:  newer vmware drivers do not work.  older (11.2.0.0) work just fine.

Win XP: works fine.

suse kde live cd seems to work, at least it will correctly resize the frame 
buffer up correctly - it doesn't boot all the way up, but this is likely due 
to my use of gcc4+tcg.