[v3,2/3] s390/vfio-ap: set status response code to 06 on gisc registration failure

Message ID 20231026183250.254432-3-akrowiak@linux.ibm.com
State New
Headers
Series a couple of corrections to the IRQ enablement function |

Commit Message

Anthony Krowiak Oct. 26, 2023, 6:32 p.m. UTC
  From: Anthony Krowiak <akrowiak@linux.ibm.com>

The interception handler for the PQAP(AQIC) command calls the
kvm_s390_gisc_register function to register the guest ISC with the channel
subsystem. If that call fails, the status response code 08 - indicating
Invalid ZONE/GISA designation - is returned to the guest. This response
code does not make sense because the non-zero return code from the
kvm_s390_gisc_register function can be due one of two things: Either the
ISC passed as a parameter by the guest to the PQAP(AQIC) command is greater
than the maximum ISC value allowed, or the guest is not using a GISA.

Since this scenario is very unlikely to happen and there is no status
response code to indicate an invalid ISC value, let's set the
response code to 06 indicating 'Invalid address of AP-queue notification
byte'. While this is not entirely accurate, it is better than indicating
that the ZONE/GISA designation is invalid which is something the guest
can do nothing about since those values are set by the hypervisor.

Signed-off-by: Anthony Krowiak <akrowiak@linux.ibm.com>
Suggested-by: Halil Pasic <pasic@linux.ibm.com>
---
 drivers/s390/crypto/vfio_ap_ops.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
  

Comments

Harald Freudenberger Oct. 27, 2023, 8:25 a.m. UTC | #1
On 2023-10-26 20:32, Tony Krowiak wrote:
> From: Anthony Krowiak <akrowiak@linux.ibm.com>
> 
> The interception handler for the PQAP(AQIC) command calls the
> kvm_s390_gisc_register function to register the guest ISC with the 
> channel
> subsystem. If that call fails, the status response code 08 - indicating
> Invalid ZONE/GISA designation - is returned to the guest. This response
> code does not make sense because the non-zero return code from the
> kvm_s390_gisc_register function can be due one of two things: Either 
> the
> ISC passed as a parameter by the guest to the PQAP(AQIC) command is 
> greater
> than the maximum ISC value allowed, or the guest is not using a GISA.
> 
> Since this scenario is very unlikely to happen and there is no status
> response code to indicate an invalid ISC value, let's set the
> response code to 06 indicating 'Invalid address of AP-queue 
> notification
> byte'. While this is not entirely accurate, it is better than 
> indicating
> that the ZONE/GISA designation is invalid which is something the guest
> can do nothing about since those values are set by the hypervisor.
> 
> Signed-off-by: Anthony Krowiak <akrowiak@linux.ibm.com>
> Suggested-by: Halil Pasic <pasic@linux.ibm.com>
> ---
>  drivers/s390/crypto/vfio_ap_ops.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/s390/crypto/vfio_ap_ops.c
> b/drivers/s390/crypto/vfio_ap_ops.c
> index 9cb28978c186..25d7ce2094f8 100644
> --- a/drivers/s390/crypto/vfio_ap_ops.c
> +++ b/drivers/s390/crypto/vfio_ap_ops.c
> @@ -393,8 +393,8 @@ static int ensure_nib_shared(unsigned long addr,
> struct gmap *gmap)
>   * Register the guest ISC to GIB interface and retrieve the
>   * host ISC to issue the host side PQAP/AQIC
>   *
> - * Response.status may be set to AP_RESPONSE_INVALID_ADDRESS in case 
> the
> - * vfio_pin_pages failed.
> + * status.response_code may be set to AP_RESPONSE_INVALID_ADDRESS in 
> case the
> + * vfio_pin_pages or kvm_s390_gisc_register failed.
>   *
>   * Otherwise return the ap_queue_status returned by the ap_aqic(),
>   * all retry handling will be done by the guest.
> @@ -458,7 +458,7 @@ static struct ap_queue_status
> vfio_ap_irq_enable(struct vfio_ap_queue *q,
>  				 __func__, nisc, isc, q->apqn);
> 
>  		vfio_unpin_pages(&q->matrix_mdev->vdev, nib, 1);
> -		status.response_code = AP_RESPONSE_INVALID_GISA;
> +		status.response_code = AP_RESPONSE_INVALID_ADDRESS;
>  		return status;
>  	}

Interesting ... The INVALID_GISA is handled in the default arm of the 
switch
in ap_queue.c but the INVALID_ADDRESS is handled as irq enablement 
failed.
So this change fits more to the current AP bus code. Thanks

Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
  
Halil Pasic Oct. 27, 2023, 11:19 a.m. UTC | #2
On Thu, 26 Oct 2023 14:32:44 -0400
Tony Krowiak <akrowiak@linux.ibm.com> wrote:

> Since this scenario is very unlikely to happen and there is no status
> response code to indicate an invalid ISC value, let's set the

Again invalid ISC won't happen except for hypervisor messes up.

> response code to 06 indicating 'Invalid address of AP-queue notification
> byte'. While this is not entirely accurate, it is better than indicating
> that the ZONE/GISA designation is invalid which is something the guest
> can do nothing about since those values are set by the hypervisor.

And more importantly AP_RESPONSE_INVALID_GISA is not valid for G2 in
the given scenario, since G2 is not trying to set up interrupts on behalf
of the G3 with a G3 GISA, but G2 is trying to set up interrupts for
itself. And then AP_RESPONSE_INVALID_GISA is architecturally simply not
a valid RC!

> 
> Signed-off-by: Anthony Krowiak <akrowiak@linux.ibm.com>
> Suggested-by: Halil Pasic <pasic@linux.ibm.com>

Except for the explanation in the commit message, the patch is good. It
is up to you if you want to fix the commit message or not.

Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
  
Anthony Krowiak Oct. 27, 2023, 3:25 p.m. UTC | #3
On 10/27/23 07:19, Halil Pasic wrote:
> On Thu, 26 Oct 2023 14:32:44 -0400
> Tony Krowiak <akrowiak@linux.ibm.com> wrote:
> 
>> Since this scenario is very unlikely to happen and there is no status
>> response code to indicate an invalid ISC value, let's set the
> 
> Again invalid ISC won't happen except for hypervisor messes up.

Again, that is one of the checks performed by the kvm_s390_gisc_register
function; however, I get your point and will remove reference in the 
comment.

> 
>> response code to 06 indicating 'Invalid address of AP-queue notification
>> byte'. While this is not entirely accurate, it is better than indicating
>> that the ZONE/GISA designation is invalid which is something the guest
>> can do nothing about since those values are set by the hypervisor.
> 
> And more importantly AP_RESPONSE_INVALID_GISA is not valid for G2 in
> the given scenario, since G2 is not trying to set up interrupts on behalf
> of the G3 with a G3 GISA, but G2 is trying to set up interrupts for
> itself. And then AP_RESPONSE_INVALID_GISA is architecturally simply not
> a valid RC!

Got it.

> 
>>
>> Signed-off-by: Anthony Krowiak <akrowiak@linux.ibm.com>
>> Suggested-by: Halil Pasic <pasic@linux.ibm.com>
> 
> Except for the explanation in the commit message, the patch is good. It
> is up to you if you want to fix the commit message or not.
> 

I'll fix the commit message.

> Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
  

Patch

diff --git a/drivers/s390/crypto/vfio_ap_ops.c b/drivers/s390/crypto/vfio_ap_ops.c
index 9cb28978c186..25d7ce2094f8 100644
--- a/drivers/s390/crypto/vfio_ap_ops.c
+++ b/drivers/s390/crypto/vfio_ap_ops.c
@@ -393,8 +393,8 @@  static int ensure_nib_shared(unsigned long addr, struct gmap *gmap)
  * Register the guest ISC to GIB interface and retrieve the
  * host ISC to issue the host side PQAP/AQIC
  *
- * Response.status may be set to AP_RESPONSE_INVALID_ADDRESS in case the
- * vfio_pin_pages failed.
+ * status.response_code may be set to AP_RESPONSE_INVALID_ADDRESS in case the
+ * vfio_pin_pages or kvm_s390_gisc_register failed.
  *
  * Otherwise return the ap_queue_status returned by the ap_aqic(),
  * all retry handling will be done by the guest.
@@ -458,7 +458,7 @@  static struct ap_queue_status vfio_ap_irq_enable(struct vfio_ap_queue *q,
 				 __func__, nisc, isc, q->apqn);
 
 		vfio_unpin_pages(&q->matrix_mdev->vdev, nib, 1);
-		status.response_code = AP_RESPONSE_INVALID_GISA;
+		status.response_code = AP_RESPONSE_INVALID_ADDRESS;
 		return status;
 	}