Linux Virtualization - May 2019 - [PATCH v2 8/8] virtio/s390: make airq summary indicators DMA

On Thu, 23 May 2019 18:22:09 +0200
Michael Mueller <mimu at linux.ibm.com> wrote:
> From: Halil Pasic <pasic at linux.ibm.com>
> 
> Hypervisor needs to interact with the summary indicators, so these
> need to be DMA memory as well (at least for protected virtualization
> guests).
> 
> Signed-off-by: Halil Pasic <pasic at linux.ibm.com>
> ---
>  drivers/s390/virtio/virtio_ccw.c | 22 +++++++++++++++-------
>  1 file changed, 15 insertions(+), 7 deletions(-)
(...)
> @@ -1501,6 +1508,7 @@ static int __init virtio_ccw_init(void)
>  {
>  	/* parse no_auto string before we do anything further */
>  	no_auto_parse();
> +	summary_indicators = cio_dma_zalloc(MAX_AIRQ_AREAS);
What happens if this fails?
>  	return ccw_driver_register(&virtio_ccw_driver);
>  }
>  device_initcall(virtio_ccw_init);

On Mon, 27 May 2019 14:00:18 +0200
Cornelia Huck <cohuck at redhat.com> wrote:
> On Thu, 23 May 2019 18:22:09 +0200
> Michael Mueller <mimu at linux.ibm.com> wrote:
> 
> > From: Halil Pasic <pasic at linux.ibm.com>
> > 
> > Hypervisor needs to interact with the summary indicators, so these
> > need to be DMA memory as well (at least for protected virtualization
> > guests).
> > 
> > Signed-off-by: Halil Pasic <pasic at linux.ibm.com>
> > ---
> >  drivers/s390/virtio/virtio_ccw.c | 22 +++++++++++++++-------
> >  1 file changed, 15 insertions(+), 7 deletions(-)
> 
> (...)
> 
> > @@ -1501,6 +1508,7 @@ static int __init virtio_ccw_init(void)
> >  {
> >  	/* parse no_auto string before we do anything further */
> >  	no_auto_parse();
> > +	summary_indicators = cio_dma_zalloc(MAX_AIRQ_AREAS);
> 
> What happens if this fails?
Bad things could happen!

How about adding

if (!summary_indicators)
	virtio_ccw_use_airq = 0; /* fall back to classic */

?

Since it ain't very likely to happen, we could also just fail
virtio_ccw_init() with -ENOMEM.

Regards,
Halil

> 
> >  	return ccw_driver_register(&virtio_ccw_driver);
> >  }
> >  device_initcall(virtio_ccw_init);
>

On Tue, 28 May 2019 16:33:42 +0200
Halil Pasic <pasic at linux.ibm.com> wrote:
> On Mon, 27 May 2019 14:00:18 +0200
> Cornelia Huck <cohuck at redhat.com> wrote:
> 
> > On Thu, 23 May 2019 18:22:09 +0200
> > Michael Mueller <mimu at linux.ibm.com> wrote:
> >   
> > > From: Halil Pasic <pasic at linux.ibm.com>
> > > 
> > > Hypervisor needs to interact with the summary indicators, so
these
> > > need to be DMA memory as well (at least for protected
virtualization
> > > guests).
> > > 
> > > Signed-off-by: Halil Pasic <pasic at linux.ibm.com>
> > > ---
> > >  drivers/s390/virtio/virtio_ccw.c | 22 +++++++++++++++-------
> > >  1 file changed, 15 insertions(+), 7 deletions(-)  
> > 
> > (...)
> >   
> > > @@ -1501,6 +1508,7 @@ static int __init virtio_ccw_init(void)
> > >  {
> > >  	/* parse no_auto string before we do anything further */
> > >  	no_auto_parse();
> > > +	summary_indicators = cio_dma_zalloc(MAX_AIRQ_AREAS);  
> > 
> > What happens if this fails?  
> 
> Bad things could happen!
> 
> How about adding
> 
> if (!summary_indicators)
> 	virtio_ccw_use_airq = 0; /* fall back to classic */
> 
> ?
> 
> Since it ain't very likely to happen, we could also just fail
> virtio_ccw_init() with -ENOMEM.
How high are the chances of things working if we fail to allocate here?
Returning with -ENOMEM is probably the more reasonable approach here.

On 28.05.19 16:33, Halil Pasic wrote:
> On Mon, 27 May 2019 14:00:18 +0200
> Cornelia Huck <cohuck at redhat.com> wrote:
> 
>> On Thu, 23 May 2019 18:22:09 +0200
>> Michael Mueller <mimu at linux.ibm.com> wrote:
>>
>>> From: Halil Pasic <pasic at linux.ibm.com>
>>>
>>> Hypervisor needs to interact with the summary indicators, so these
>>> need to be DMA memory as well (at least for protected
virtualization
>>> guests).
>>>
>>> Signed-off-by: Halil Pasic <pasic at linux.ibm.com>
>>> ---
>>>   drivers/s390/virtio/virtio_ccw.c | 22 +++++++++++++++-------
>>>   1 file changed, 15 insertions(+), 7 deletions(-)
>>
>> (...)
>>
>>> @@ -1501,6 +1508,7 @@ static int __init virtio_ccw_init(void)
>>>   {
>>>   	/* parse no_auto string before we do anything further */
>>>   	no_auto_parse();
>>> +	summary_indicators = cio_dma_zalloc(MAX_AIRQ_AREAS);
>>
>> What happens if this fails?
> 
> Bad things could happen!
> 
> How about adding
> 
> if (!summary_indicators)
> 	virtio_ccw_use_airq = 0; /* fall back to classic */
> 
> ?
> 
> Since it ain't very likely to happen, we could also just fail
> virtio_ccw_init() with -ENOMEM.
That is what I'm currently doing in v3.
> 
> Regards,
> Halil
> 
> 
>>
>>>   	return ccw_driver_register(&virtio_ccw_driver);
>>>   }
>>>   device_initcall(virtio_ccw_init);
>>
> 
Michael