Re: [Qemu-devel] [PATCH for-2.4] Revert "vhost-user: Send VHOST_RESET_OWNER on vhost stop"

[Michael S. Tsirkin]

On Tue, Aug 04, 2015 at 09:00:47PM +0200, Luke Gorrie wrote:
> Hi Michael,
> 
> Sorry I didn't see this mail sooner -
> 
> On 30 July 2015 at 10:36, Michael S. Tsirkin <address@hidden> wrote:
> 
>     This reverts commit 294ce717e0f212ed0763307f3eab72b4a1bdf4d0.
> 
>     vhost stop happens e.g. when guest unloads the driver,
>     so closing the backend connection is not the right
>     thing to do here.
> 
>     VHOST_RESET_OWNER should happen on vhost_dev_cleanup - it's
>     the counterpart of VHOST_SET_OWNER.
> 
>     Cc: Luke Gorrie <address@hidden>
>     Signed-off-by: Michael S. Tsirkin <address@hidden>
>     ---
> 
>     I think we need this in 2.4 to avoid introducing regressions
>     in the protocol. We'll fix properly in 2.5.
>     Luke, can you comment please?
> 
> 
> Interesting. Currently we don't have a test case in Snabb Switch CI for guests
> reloading drivers. We need to add that and let you know how it goes.
> 
> The reason I made this patch was to fix a problem with guests failing to
> reboot. This patch seemed to be necessary for the vswitch to know that the
> vring in guest memory is not valid anymore and should not be used for DMA
> (since that memory is being reallocated to who knows what).

This makes sense, but the description says master will
close the socket. This makes no sense to me.
Maybe we should leave the code as is, but rename the message
to RESET_DEVICE then?

> I wonder if driver reload sort of behavior can be related to this interesting
> problem that we are also tracking down - ideas would be very welcome:
> https://github.com/SnabbCo/snabbswitch/issues/574
> 
>