The qdisk implementation is using the native xenbus protocol only in
case of no protocol specified at all. As using the explicit 32- or
64-bit protocol is slower than the native one due to copying requests
not by memcpy but element for element, this is not optimal.
Correct this by using the native
On 05/07/16 16:55, Stefano Stabellini wrote:
> On Wed, 29 Jun 2016, Anthony PERARD wrote:
>> On Wed, Jun 29, 2016 at 05:50:48PM +0200, Juergen Gross wrote:
>>> The qdisk implementation is using the native xenbus protocol only in
>>> case of no protocol specified at all. As using the explicit 32- or
On Tue, 30 Aug 2016, Juergen Gross wrote:
> On 05/07/16 16:55, Stefano Stabellini wrote:
> > On Wed, 29 Jun 2016, Anthony PERARD wrote:
> >> On Wed, Jun 29, 2016 at 05:50:48PM +0200, Juergen Gross wrote:
> >>> The qdisk implementation is using the native xenbus protocol only in
> >>> case of no pro
On Wed, Jun 29, 2016 at 05:50:48PM +0200, Juergen Gross wrote:
> The qdisk implementation is using the native xenbus protocol only in
> case of no protocol specified at all. As using the explicit 32- or
> 64-bit protocol is slower than the native one due to copying requests
> not by memcpy but elem
On Wed, 29 Jun 2016, Anthony PERARD wrote:
> On Wed, Jun 29, 2016 at 05:50:48PM +0200, Juergen Gross wrote:
> > The qdisk implementation is using the native xenbus protocol only in
> > case of no protocol specified at all. As using the explicit 32- or
> > 64-bit protocol is slower than the native o
