[RFC,v1,4/4] vfio-ccw: Don't call cp_free if we are processing a channel program
diff mbox series

Message ID 31c3c29e3e9c4f0312f9363a1c3a5d22b74f68cb.1561997809.git.alifm@linux.ibm.com
State New
Headers show
Series
  • Some vfio-ccw fixes
Related show

Commit Message

Farhan Ali July 1, 2019, 4:23 p.m. UTC
There is a small window where it's possible that we could be working
on an interrupt (queued in the workqueue) and setting up a channel
program (i.e allocating memory, pinning pages, translating address).
This can lead to allocating and freeing the channel program at the
same time and can cause memory corruption.

Let's not call cp_free if we are currently processing a channel program.
The only way we know for sure that we don't have a thread setting
up a channel program is when the state is set to VFIO_CCW_STATE_CP_PENDING.

Signed-off-by: Farhan Ali <alifm@linux.ibm.com>
---
 drivers/s390/cio/vfio_ccw_drv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Cornelia Huck July 2, 2019, 9:51 a.m. UTC | #1
On Mon,  1 Jul 2019 12:23:46 -0400
Farhan Ali <alifm@linux.ibm.com> wrote:

> There is a small window where it's possible that we could be working
> on an interrupt (queued in the workqueue) and setting up a channel
> program (i.e allocating memory, pinning pages, translating address).
> This can lead to allocating and freeing the channel program at the
> same time and can cause memory corruption.

This can only happen if the interrupt is for a halt/clear operation,
right?

> 
> Let's not call cp_free if we are currently processing a channel program.
> The only way we know for sure that we don't have a thread setting
> up a channel program is when the state is set to VFIO_CCW_STATE_CP_PENDING.

I have looked through the code again and I think you are right.

> 
> Signed-off-by: Farhan Ali <alifm@linux.ibm.com>
> ---
>  drivers/s390/cio/vfio_ccw_drv.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/s390/cio/vfio_ccw_drv.c b/drivers/s390/cio/vfio_ccw_drv.c
> index 4e3a903..0357165 100644
> --- a/drivers/s390/cio/vfio_ccw_drv.c
> +++ b/drivers/s390/cio/vfio_ccw_drv.c
> @@ -92,7 +92,7 @@ static void vfio_ccw_sch_io_todo(struct work_struct *work)
>  		     (SCSW_ACTL_DEVACT | SCSW_ACTL_SCHACT));
>  	if (scsw_is_solicited(&irb->scsw)) {
>  		cp_update_scsw(&private->cp, &irb->scsw);
> -		if (is_final)
> +		if (is_final && private->state == VFIO_CCW_STATE_CP_PENDING)

Do we actually want to call cp_update_scsw() unconditionally?

At this point, we know that we have a solicited interrupt; that may be
for several reasons:
- Interrupt for something we issued via ssch; it makes sense to update
  the scsw with the cpa address.
- Interrupt for a csch; the cpa address will be unpredictable, even if
  we did a ssch before. cp_update_scsw() hopefully can deal with that?
  Given that its purpose is to translate the cpa back, any
  unpredictable value in the scsw should be fine in the end.
- Interrupt for a hsch after we did a ssch; the cpa might be valid (see
  figure 16-6).
- Interrupt for a hsch without a prior ssch; we'll end up with an
  unpredictable cpa, again.

So I *think* we're fine with calling cp_update_scsw() in all cases,
even if there's junk in the cpa of the scsw we get from the hardware.
Opinions?

>  			cp_free(&private->cp);
>  	}
>  	mutex_lock(&private->io_mutex);

Patch
diff mbox series

diff --git a/drivers/s390/cio/vfio_ccw_drv.c b/drivers/s390/cio/vfio_ccw_drv.c
index 4e3a903..0357165 100644
--- a/drivers/s390/cio/vfio_ccw_drv.c
+++ b/drivers/s390/cio/vfio_ccw_drv.c
@@ -92,7 +92,7 @@  static void vfio_ccw_sch_io_todo(struct work_struct *work)
 		     (SCSW_ACTL_DEVACT | SCSW_ACTL_SCHACT));
 	if (scsw_is_solicited(&irb->scsw)) {
 		cp_update_scsw(&private->cp, &irb->scsw);
-		if (is_final)
+		if (is_final && private->state == VFIO_CCW_STATE_CP_PENDING)
 			cp_free(&private->cp);
 	}
 	mutex_lock(&private->io_mutex);