[PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Ian Jackson]

[ extra patch, following test report from Ross ]

Ross reports that stdvga guests do not work, and cirrus guests are
slow, because qemu tries to do xc_domain_add_to_physmap.  We will need
another dmop to fix this properly.

For now, document the problem.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
Reported-by: Ross Lagerwall <ross.lagerwall@citrix.com>
CC: Ross Lagerwall <ross.lagerwall@citrix.com>
CC: Wei Liu <wei.liu2@citrix.com>
CC: Paul Durrant <Paul.Durrant@citrix.com>
---
 docs/man/xl.cfg.pod.5.in | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/docs/man/xl.cfg.pod.5.in b/docs/man/xl.cfg.pod.5.in
index 8446d43..46b5ae6 100644
--- a/docs/man/xl.cfg.pod.5.in
+++ b/docs/man/xl.cfg.pod.5.in
@@ -2239,6 +2239,12 @@ For example, cdrom insert will fail.
 
 =item
 
+Domains with stdvga graphics cards to not work.
+Domains with cirrus vga do work,
+but there are performance problems with the vga console.
+
+=item
+
 You must create user(s) for qemu to run as.
 
 Ideally, set aside a range of 32752 uids
-- 
2.1.4


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

Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Wei Liu]

On Fri, Oct 06, 2017 at 03:33:52PM +0100, Ian Jackson wrote:
> [ extra patch, following test report from Ross ]
> 
> Ross reports that stdvga guests do not work, and cirrus guests are
> slow, because qemu tries to do xc_domain_add_to_physmap.  We will need
> another dmop to fix this properly.
> 
> For now, document the problem.
> 
> Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
> Reported-by: Ross Lagerwall <ross.lagerwall@citrix.com>
> CC: Ross Lagerwall <ross.lagerwall@citrix.com>
> CC: Wei Liu <wei.liu2@citrix.com>
> CC: Paul Durrant <Paul.Durrant@citrix.com>
> ---
>  docs/man/xl.cfg.pod.5.in | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/docs/man/xl.cfg.pod.5.in b/docs/man/xl.cfg.pod.5.in
> index 8446d43..46b5ae6 100644
> --- a/docs/man/xl.cfg.pod.5.in
> +++ b/docs/man/xl.cfg.pod.5.in
> @@ -2239,6 +2239,12 @@ For example, cdrom insert will fail.
>  
>  =item
>  
> +Domains with stdvga graphics cards to not work.
> +Domains with cirrus vga do work,
> +but there are performance problems with the vga console.
> +

I don't follow. Do you imply cirrus doesn't need add_to_physmap to
function?

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

Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Ross Lagerwall]

On 10/06/2017 03:51 PM, Wei Liu wrote:
> On Fri, Oct 06, 2017 at 03:33:52PM +0100, Ian Jackson wrote:
>> [ extra patch, following test report from Ross ]
>>
>> Ross reports that stdvga guests do not work, and cirrus guests are
>> slow, because qemu tries to do xc_domain_add_to_physmap.  We will need
>> another dmop to fix this properly.
>>
>> For now, document the problem.
>>
>> Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
>> Reported-by: Ross Lagerwall <ross.lagerwall@citrix.com>
>> CC: Ross Lagerwall <ross.lagerwall@citrix.com>
>> CC: Wei Liu <wei.liu2@citrix.com>
>> CC: Paul Durrant <Paul.Durrant@citrix.com>
>> ---
>>   docs/man/xl.cfg.pod.5.in | 6 ++++++
>>   1 file changed, 6 insertions(+)
>>
>> diff --git a/docs/man/xl.cfg.pod.5.in b/docs/man/xl.cfg.pod.5.in
>> index 8446d43..46b5ae6 100644
>> --- a/docs/man/xl.cfg.pod.5.in
>> +++ b/docs/man/xl.cfg.pod.5.in
>> @@ -2239,6 +2239,12 @@ For example, cdrom insert will fail.
>>   
>>   =item
>>   
>> +Domains with stdvga graphics cards to not work.
>> +Domains with cirrus vga do work,
>> +but there are performance problems with the vga console.
>> +
> 
> I don't follow. Do you imply cirrus doesn't need add_to_physmap to
> function?
> 

For both stdvga and cirrus, the function gets called and fails (although 
QEMU doesn't log an error by default). With stdvga the VM wedged. With 
cirrus, the VM worked but the display performance seemed worse than normal.

-- 
Ross Lagerwall

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

Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Paul Durrant]

> -----Original Message-----
> From: Wei Liu [mailto:wei.liu2@citrix.com]
> Sent: 06 October 2017 15:51
> To: Ian Jackson <Ian.Jackson@citrix.com>
> Cc: xen-devel@lists.xensource.com; Ross Lagerwall
> <ross.lagerwall@citrix.com>; Wei Liu <wei.liu2@citrix.com>; Paul Durrant
> <Paul.Durrant@citrix.com>
> Subject: Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack
> of physmap dmop
> 
> On Fri, Oct 06, 2017 at 03:33:52PM +0100, Ian Jackson wrote:
> > [ extra patch, following test report from Ross ]
> >
> > Ross reports that stdvga guests do not work, and cirrus guests are
> > slow, because qemu tries to do xc_domain_add_to_physmap.  We will
> need
> > another dmop to fix this properly.
> >
> > For now, document the problem.
> >
> > Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
> > Reported-by: Ross Lagerwall <ross.lagerwall@citrix.com>
> > CC: Ross Lagerwall <ross.lagerwall@citrix.com>
> > CC: Wei Liu <wei.liu2@citrix.com>
> > CC: Paul Durrant <Paul.Durrant@citrix.com>
> > ---
> >  docs/man/xl.cfg.pod.5.in | 6 ++++++
> >  1 file changed, 6 insertions(+)
> >
> > diff --git a/docs/man/xl.cfg.pod.5.in b/docs/man/xl.cfg.pod.5.in
> > index 8446d43..46b5ae6 100644
> > --- a/docs/man/xl.cfg.pod.5.in
> > +++ b/docs/man/xl.cfg.pod.5.in
> > @@ -2239,6 +2239,12 @@ For example, cdrom insert will fail.
> >
> >  =item
> >
> > +Domains with stdvga graphics cards to not work.
> > +Domains with cirrus vga do work,
> > +but there are performance problems with the vga console.
> > +
> 
> I don't follow. Do you imply cirrus doesn't need add_to_physmap to
> function?

I suspect what it happening is that the VRAM is not moved into the BAR, but the emulation in QEMU instead handles the BAR writes and then updates the VRAM in its current location.

  Paul

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

Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Ian Jackson]

Wei Liu writes ("Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop"):
> On Fri, Oct 06, 2017 at 03:33:52PM +0100, Ian Jackson wrote:
> > [ extra patch, following test report from Ross ]
...
> > +Domains with stdvga graphics cards to not work.
> > +Domains with cirrus vga do work,
> > +but there are performance problems with the vga console.
> > +
> 
> I don't follow. Do you imply cirrus doesn't need add_to_physmap to
> function?

Evidently not.  My qemu-dm-debian.guest.osstest.log does not contain
any of the errors Ross quotes and I can connect to my guest's VGA
console over vnc.

It does take a little extra time to boot, I think.  I didn't really
notice this and thought it was a side effect of the happenstance that
my guest had been converted from PV...

Ian.

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

Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Wei Liu]

On Fri, Oct 06, 2017 at 03:57:09PM +0100, Ian Jackson wrote:
> Wei Liu writes ("Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop"):
> > On Fri, Oct 06, 2017 at 03:33:52PM +0100, Ian Jackson wrote:
> > > [ extra patch, following test report from Ross ]
> ...
> > > +Domains with stdvga graphics cards to not work.
> > > +Domains with cirrus vga do work,
> > > +but there are performance problems with the vga console.
> > > +
> > 
> > I don't follow. Do you imply cirrus doesn't need add_to_physmap to
> > function?
> 
> Evidently not.  My qemu-dm-debian.guest.osstest.log does not contain
> any of the errors Ross quotes and I can connect to my guest's VGA
> console over vnc.
> 
> It does take a little extra time to boot, I think.  I didn't really
> notice this and thought it was a side effect of the happenstance that
> my guest had been converted from PV...
> 

Let's just say they don't work properly yet?

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

Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Ian Jackson]

Paul Durrant writes ("RE: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop"):
> I suspect what it happening is that the VRAM is not moved into the BAR, but the emulation in QEMU instead handles the BAR writes and then updates the VRAM in its current location.

Is it going to be a problem that the vram is still mapped in its
previous location ?

Ian.

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

Re: [PATCH v2 25/24] xl: Document VGA problems arising from lack of physmap dmop

[Paul Durrant]

> -----Original Message-----
> From: Ian Jackson [mailto:ian.jackson@eu.citrix.com]
> Sent: 06 October 2017 16:02
> To: Paul Durrant <Paul.Durrant@citrix.com>
> Cc: Wei Liu <wei.liu2@citrix.com>; xen-devel@lists.xensource.com; Ross
> Lagerwall <ross.lagerwall@citrix.com>
> Subject: RE: [PATCH v2 25/24] xl: Document VGA problems arising from lack
> of physmap dmop
> 
> Paul Durrant writes ("RE: [PATCH v2 25/24] xl: Document VGA problems
> arising from lack of physmap dmop"):
> > I suspect what it happening is that the VRAM is not moved into the BAR,
> but the emulation in QEMU instead handles the BAR writes and then updates
> the VRAM in its current location.
> 
> Is it going to be a problem that the vram is still mapped in its
> previous location ?
> 

Unclear. I think it of those things where QEMU knows where it is but nothing else really does. With trad the VRAM was always just below the 4G boundary in guest space but this is not the case with upstream so I don't know what the risk of something else trying to put memory (or a hotplugged device) in the same space is.

  Paul

> Ian.

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