Re: [PATCH v2] rpmsg: virtio: add endianness conversions

[Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>]

On Wed, Jul 01, 2020 at 06:00:30AM -0400, Michael S. Tsirkin wrote:
> On Tue, Jun 30, 2020 at 11:50:57AM -0600, Mathieu Poirier wrote:
> > On Tue, Jun 16, 2020 at 09:30:11AM +0200, Guennadi Liakhovetski wrote:
> > > According to the VirtIO 1.0 spec data, sent over virtual queues must
> > > be in little-endian format. Update the RPMsg VirtIO implementation
> > > to enforce that but let legacy configurations continue use native
> > > endianness.
> > > 
> > > Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
> > > ---
> > > 
> > > v2: Following suggestions from Michael and Mathieu switch to using virtio16/32 
> > > types and conversion functions.
> > > 
> > >  drivers/rpmsg/virtio_rpmsg_bus.c | 63 ++++++++++++++++++++++------------------
> > >  1 file changed, 35 insertions(+), 28 deletions(-)
> > > 
> > > diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c b/drivers/rpmsg/virtio_rpmsg_bus.c
> > > index 07d4f33..41025df 100644
> > > --- a/drivers/rpmsg/virtio_rpmsg_bus.c
> > > +++ b/drivers/rpmsg/virtio_rpmsg_bus.c
> > > @@ -11,6 +11,8 @@
> > >  
> > >  #define pr_fmt(fmt) "%s: " fmt, __func__
> > >  
> > > +#include <asm/byteorder.h>
> > > +
> > 
> > Not sure about this header - what does it provide?

It was needed for the original cpu_to_le... and friends version of the patch.

> > 
> > >  #include <linux/dma-mapping.h>
> > >  #include <linux/idr.h>
> > >  #include <linux/jiffies.h>
> > > @@ -22,6 +24,7 @@
> > >  #include <linux/scatterlist.h>
> > >  #include <linux/slab.h>
> > >  #include <linux/sched.h>
> > > +#include <linux/types.h>
> > 
> > This too... It should be fine with linux/virtio_config.h below since it 
> > already includes linux/virtio_byteorder.h
> 
> It's probably best to include linux/virtio_byteorder.h explicitly.

Yep, will do.

> > Moreover, please send another revision that applies on rproc-next.  The third
> > hunk in this patch doesn't apply.
> > 
> > With the above:
> > 
> > Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
> > 
> > >  #include <linux/virtio.h>
> > >  #include <linux/virtio_ids.h>
> > >  #include <linux/virtio_config.h>

[snip]

> > > @@ -846,15 +853,15 @@ static int rpmsg_ns_cb(struct rpmsg_device *rpdev, void *data, int len,
> > >  	/* don't trust the remote processor for null terminating the name */
> > >  	msg->name[RPMSG_NAME_SIZE - 1] = '\0';
> > >  
> > > -	dev_info(dev, "%sing channel %s addr 0x%x\n",
> > > -		 msg->flags & RPMSG_NS_DESTROY ? "destroy" : "creat",
> > > -		 msg->name, msg->addr);
> > > -
> > >  	strncpy(chinfo.name, msg->name, sizeof(chinfo.name));
> > >  	chinfo.src = RPMSG_ADDR_ANY;
> > > -	chinfo.dst = msg->addr;
> > > +	chinfo.dst = virtio32_to_cpu(vrp->vdev, msg->addr);
> > > +
> > > +	dev_info(dev, "%sing channel %s addr 0x%x\n",
> 
> 
> Let's not try tricks. Just %s and destroying/creating below.
> Wastes 3 bytes of kernel space but is clearer.

Those aren't my tricks - they're already there, but sure, I can replace that while at it.

Thanks
Guennadi

> 
> > > +		 virtio32_to_cpu(vrp->vdev, msg->flags) & RPMSG_NS_DESTROY ?
> > > +		 "destroy" : "creat", msg->name, chinfo.dst);
> > >  
> > > -	if (msg->flags & RPMSG_NS_DESTROY) {
> > > +	if (virtio32_to_cpu(vrp->vdev, msg->flags) & RPMSG_NS_DESTROY) {
> > >  		ret = rpmsg_unregister_device(&vrp->vdev->dev, &chinfo);
> > >  		if (ret)
> > >  			dev_err(dev, "rpmsg_destroy_channel failed: %d\n", ret);
> > > -- 
> > > 1.9.3
> > > 
>