Re: [virtio] [PATCH v5] virtio-blk: restore VIRTIO_BLK_F_FLUSH and VIRTIO_BLK_F_CONFIG_WCE

[Cornelia Huck <cornelia.huck@de.ibm.com>]

On Tue, 11 Aug 2015 16:47:05 +0300
"Michael S. Tsirkin" <mst@redhat.com> wrote:

> On Tue, Aug 11, 2015 at 03:41:20PM +0200, Cornelia Huck wrote:
> > On Tue, 11 Aug 2015 16:23:39 +0300
> > "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > 
> > > OK to summarize, how about:
> > > 
> > > In free text:
> > > 
> > > The cache mode after reset can be either writeback or writethrough.
> > > The actual mode can be determined by reading \field{writeback}.
> > > 
> > > In conformance paragraph:
> > > 
> > > The driver MUST NOT read \field{writeback} before setting
> > > the FEATURES_OK \field{status} bit.
> > > 
> > > 
> > > And in legacy section:
> > > 
> > > Legacy drivers and drivers did not support the FEATURES_OK \field{status} bit,
> > > and used the device before setting the DRIVER_OK bit.
> > > Therefore:
> > > 
> > > When using the legacy interface, the driver MAY read \field{writeback}
> > > before setting the FEATURES_OK \field{status} field.
> > 
> > But if it is using the legacy interface, it won't ever set FEATURES_OK,
> > no? The wording is a bit confusing.
> > 
> > "As the legacy interface did not support FEATURES_OK to signal end of
> > feature negotation, the driver MAY read \field{writeback} without the
> > FEATURES_OK \field{status} bit having been set."
> > 
> > ?
> 
> Right.
> 
> > Do we need to mandate that the device MUST assume the driver is operating
> > in legacy mode in that case?
> 
> Not sure I understand - in what case?

If it reads writeback without FEATURES_OK having been set.

> 
> > > When using the legacy interface, the transitional devices MAY assume
> > > that driver features won't be changed after reading \field{writeback}.
> > 
> > "For a driver using the legacy interface, the transitional device MAY
> > assume that the driver won't change features after reading
> > \field{writeback}."
> > 
> > ?
> 
> Except we say "When using the legacy interface" everywhere,
> let's keep it consistent pls.
> 
> 	When using the legacy interface, the transitional device MAY
> 	assume that the driver won't change features after reading
> 	\field{writeback}."
> 
> point is both driver and device are using the legacy interface.

Fine with me.