From: Mathieu Poirier <mathieu.poirier@linaro.org>
To: Arnaud POULIQUEN <arnaud.pouliquen@st.com>
Cc: "ohad@wizery.com" <ohad@wizery.com>,
"bjorn.andersson@linaro.org" <bjorn.andersson@linaro.org>,
"guennadi.liakhovetski@linux.intel.com"
<guennadi.liakhovetski@linux.intel.com>,
"linux-remoteproc@vger.kernel.org"
<linux-remoteproc@vger.kernel.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v2 3/9] rpmsg: virtio: Move from virtio to rpmsg byte conversion
Date: Thu, 15 Oct 2020 13:46:20 -0600 [thread overview]
Message-ID: <20201015194620.GE1450102@xps15> (raw)
In-Reply-To: <0264d2f6-22c7-6c6d-0db4-05fd9e0b1121@st.com>
On Wed, Oct 14, 2020 at 07:04:32PM +0200, Arnaud POULIQUEN wrote:
>
>
> On 10/14/20 1:25 AM, Mathieu Poirier wrote:
> > Use rpmsg byte conversion functions in order for the RPMSG
> > headers and generic functions to be used by external entities.
> >
> > Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
> > ---
> > drivers/rpmsg/virtio_rpmsg_bus.c | 60 +++++++++++++++++++-------------
> > 1 file changed, 35 insertions(+), 25 deletions(-)
> >
> > diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c b/drivers/rpmsg/virtio_rpmsg_bus.c
> > index 9006fc7f73d0..793fe924671f 100644
> > --- a/drivers/rpmsg/virtio_rpmsg_bus.c
> > +++ b/drivers/rpmsg/virtio_rpmsg_bus.c
> > @@ -19,11 +19,11 @@
> > #include <linux/mutex.h>
> > #include <linux/of_device.h>
> > #include <linux/rpmsg.h>
> > +#include <linux/rpmsg_byteorder.h>
> > #include <linux/scatterlist.h>
> > #include <linux/slab.h>
> > #include <linux/sched.h>
> > #include <linux/virtio.h>
> > -#include <linux/virtio_byteorder.h>
> > #include <linux/virtio_ids.h>
> > #include <linux/virtio_config.h>
> > #include <linux/wait.h>
> > @@ -85,11 +85,11 @@ struct virtproc_info {
> > * Every message sent(/received) on the rpmsg bus begins with this header.
> > */
> > struct rpmsg_hdr {
> > - __virtio32 src;
> > - __virtio32 dst;
> > - __virtio32 reserved;
> > - __virtio16 len;
> > - __virtio16 flags;
> > + __rpmsg32 src;
> > + __rpmsg32 dst;
> > + __rpmsg32 reserved;
> > + __rpmsg16 len;
> > + __rpmsg16 flags;
> > u8 data[];
> > } __packed;
> >
> > @@ -107,8 +107,8 @@ struct rpmsg_hdr {
> > */
> > struct rpmsg_ns_msg {
> > char name[RPMSG_NAME_SIZE];
> > - __virtio32 addr;
> > - __virtio32 flags;
> > + __rpmsg32 addr;
> > + __rpmsg32 flags;
> > } __packed;
> >
> > /**
> > @@ -280,6 +280,14 @@ static struct rpmsg_endpoint *__rpmsg_create_ept(struct virtproc_info *vrp,
> > return NULL;
> > }
> >
> > +static bool virtio_rpmsg_is_little_endian(struct rpmsg_device *rpdev)
> > +{
> > + struct virtio_rpmsg_channel *vch = to_virtio_rpmsg_channel(rpdev);
> > + struct virtproc_info *vrp = vch->vrp;
> > +
> > + return virtio_is_little_endian(vrp->vdev);
> > +}
>
> Regarding this i wonder if the endianess could not be a rpmsg_device field that would be set on
> __rpmsg_create_channel?
> I don't think that the endianess could change, so perhaps no need to call ops for each conversion
> using interface implemented in rpmsg.h...
That's a valid point - I certainly don't expect the endianness to change
dynamically.
> But perhaps I missed something?
>
> > +
> > static struct rpmsg_endpoint *virtio_rpmsg_create_ept(struct rpmsg_device *rpdev,
> > rpmsg_rx_cb_t cb,
> > void *priv,
> > @@ -336,8 +344,8 @@ static int virtio_rpmsg_announce_create(struct rpmsg_device *rpdev)
> > struct rpmsg_ns_msg nsm;
> >
> > strncpy(nsm.name, rpdev->id.name, RPMSG_NAME_SIZE);
> > - nsm.addr = cpu_to_virtio32(vrp->vdev, rpdev->ept->addr);
> > - nsm.flags = cpu_to_virtio32(vrp->vdev, RPMSG_NS_CREATE);
> > + nsm.addr = cpu_to_rpmsg32(rpdev, rpdev->ept->addr);
> > + nsm.flags = cpu_to_rpmsg32(rpdev, RPMSG_NS_CREATE);
> >
> > err = rpmsg_sendto(rpdev->ept, &nsm, sizeof(nsm), RPMSG_NS_ADDR);
> > if (err)
> > @@ -360,8 +368,8 @@ static int virtio_rpmsg_announce_destroy(struct rpmsg_device *rpdev)
> > struct rpmsg_ns_msg nsm;
> >
> > strncpy(nsm.name, rpdev->id.name, RPMSG_NAME_SIZE);
> > - nsm.addr = cpu_to_virtio32(vrp->vdev, rpdev->ept->addr);
> > - nsm.flags = cpu_to_virtio32(vrp->vdev, RPMSG_NS_DESTROY);
> > + nsm.addr = cpu_to_rpmsg32(rpdev, rpdev->ept->addr);
> > + nsm.flags = cpu_to_rpmsg32(rpdev, RPMSG_NS_DESTROY);
> >
> > err = rpmsg_sendto(rpdev->ept, &nsm, sizeof(nsm), RPMSG_NS_ADDR);
> > if (err)
> > @@ -372,6 +380,7 @@ static int virtio_rpmsg_announce_destroy(struct rpmsg_device *rpdev)
> > }
> >
> > static const struct rpmsg_device_ops virtio_rpmsg_ops = {
> > + .is_little_endian = virtio_rpmsg_is_little_endian,
> > .create_ept = virtio_rpmsg_create_ept,
> > .announce_create = virtio_rpmsg_announce_create,
> > .announce_destroy = virtio_rpmsg_announce_destroy,
> > @@ -613,10 +622,10 @@ static int rpmsg_send_offchannel_raw(struct rpmsg_device *rpdev,
> > }
> > }
> >
> > - msg->len = cpu_to_virtio16(vrp->vdev, len);
> > + msg->len = cpu_to_rpmsg16(rpdev, len);
> > msg->flags = 0;
> > - msg->src = cpu_to_virtio32(vrp->vdev, src);
> > - msg->dst = cpu_to_virtio32(vrp->vdev, dst);
> > + msg->src = cpu_to_rpmsg32(rpdev, src);
> > + msg->dst = cpu_to_rpmsg32(rpdev, dst);
> > msg->reserved = 0;
> > memcpy(msg->data, data, len);
> >
> > @@ -705,14 +714,15 @@ static int rpmsg_recv_single(struct virtproc_info *vrp, struct device *dev,
> > {
> > struct rpmsg_endpoint *ept;
> > struct scatterlist sg;
> > - unsigned int msg_len = virtio16_to_cpu(vrp->vdev, msg->len);
> > + bool little_endian = rpmsg_is_little_endian();
> > + unsigned int msg_len = __rpmsg16_to_cpu(little_endian, msg->len);
> > int err;
> >
> > dev_dbg(dev, "From: 0x%x, To: 0x%x, Len: %d, Flags: %d, Reserved: %d\n",
> > - virtio32_to_cpu(vrp->vdev, msg->src),
> > - virtio32_to_cpu(vrp->vdev, msg->dst), msg_len,
> > - virtio16_to_cpu(vrp->vdev, msg->flags),
> > - virtio32_to_cpu(vrp->vdev, msg->reserved));
> > + __rpmsg32_to_cpu(little_endian, msg->src),
> > + __rpmsg32_to_cpu(little_endian, msg->dst), msg_len,
> > + __rpmsg16_to_cpu(little_endian, msg->flags),
> > + __rpmsg32_to_cpu(little_endian, msg->reserved));
>
> Nitpicking: sometime rpmsgXX_to_cpu is used, sometime __rpmsgXX_to_cpu,
> Perhaps only one API should be used... But i don't see any blocking point to use both...:)
Here we have to use __rpmsgXX_to_CPU() because the rpmsg_device is not
available. It could be fixed but would require a serious amount of refactoring
for little value. Moreover since this rpmsg_recv_single() is specific to the
virtIO implementation the end result is the same.
>
> Thanks,
> Arnaud
>
> > #if defined(CONFIG_DYNAMIC_DEBUG)
> > dynamic_hex_dump("rpmsg_virtio RX: ", DUMP_PREFIX_NONE, 16, 1,
> > msg, sizeof(*msg) + msg_len, true);
> > @@ -731,7 +741,7 @@ static int rpmsg_recv_single(struct virtproc_info *vrp, struct device *dev,
> > /* use the dst addr to fetch the callback of the appropriate user */
> > mutex_lock(&vrp->endpoints_lock);
> >
> > - ept = idr_find(&vrp->endpoints, virtio32_to_cpu(vrp->vdev, msg->dst));
> > + ept = idr_find(&vrp->endpoints, __rpmsg32_to_cpu(little_endian, msg->dst));
> >
> > /* let's make sure no one deallocates ept while we use it */
> > if (ept)
> > @@ -745,7 +755,7 @@ static int rpmsg_recv_single(struct virtproc_info *vrp, struct device *dev,
> >
> > if (ept->cb)
> > ept->cb(ept->rpdev, msg->data, msg_len, ept->priv,
> > - virtio32_to_cpu(vrp->vdev, msg->src));
> > + __rpmsg32_to_cpu(little_endian, msg->src));
> >
> > mutex_unlock(&ept->cb_lock);
> >
> > @@ -853,13 +863,13 @@ static int rpmsg_ns_cb(struct rpmsg_device *rpdev, void *data, int len,
> >
> > strncpy(chinfo.name, msg->name, sizeof(chinfo.name));
> > chinfo.src = RPMSG_ADDR_ANY;
> > - chinfo.dst = virtio32_to_cpu(vrp->vdev, msg->addr);
> > + chinfo.dst = rpmsg32_to_cpu(rpdev, msg->addr);
> >
> > dev_info(dev, "%sing channel %s addr 0x%x\n",
> > - virtio32_to_cpu(vrp->vdev, msg->flags) & RPMSG_NS_DESTROY ?
> > + rpmsg32_to_cpu(rpdev, msg->flags) & RPMSG_NS_DESTROY ?
> > "destroy" : "creat", msg->name, chinfo.dst);
> >
> > - if (virtio32_to_cpu(vrp->vdev, msg->flags) & RPMSG_NS_DESTROY) {
> > + if (rpmsg32_to_cpu(rpdev, 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);
> >
next prev parent reply other threads:[~2020-10-15 19:46 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-13 23:25 [PATCH v2 0/9] rpmsg: Make RPMSG name service modular Mathieu Poirier
2020-10-13 23:25 ` [PATCH v2 1/9] rpmsg: Move rpmsg_endpoint_ops to rpmsg.h Mathieu Poirier
2020-10-15 8:48 ` Arnaud POULIQUEN
2020-10-13 23:25 ` [PATCH v2 2/9] rpmsg: Introduce __rpmsg{16|32|64} types Mathieu Poirier
2020-10-14 16:31 ` Arnaud POULIQUEN
2020-10-15 19:32 ` Mathieu Poirier
2020-10-16 9:41 ` Arnaud POULIQUEN
2020-10-13 23:25 ` [PATCH v2 3/9] rpmsg: virtio: Move from virtio to rpmsg byte conversion Mathieu Poirier
2020-10-14 17:04 ` Arnaud POULIQUEN
2020-10-15 19:46 ` Mathieu Poirier [this message]
2020-10-13 23:25 ` [PATCH v2 4/9] rpmsg: Move rpmsg_hr and rpmsg_ns_msg to header file Mathieu Poirier
2020-10-15 8:33 ` Arnaud POULIQUEN
2020-10-15 20:19 ` Mathieu Poirier
2020-10-16 9:45 ` Arnaud POULIQUEN
2020-10-13 23:25 ` [PATCH v2 5/9] rpmsg: virtio: Rename rpmsg_create_channel Mathieu Poirier
2020-10-13 23:25 ` [PATCH v2 6/9] rpmsg: core: Add channel creation internal API Mathieu Poirier
2020-10-13 23:25 ` [PATCH v2 7/9] rpmsg: virtio: Add rpmsg channel device ops Mathieu Poirier
2020-10-13 23:25 ` [PATCH v2 8/9] rpmsg: Make rpmsg_{register|unregister}_device() public Mathieu Poirier
2020-10-15 8:47 ` Arnaud POULIQUEN
2020-10-13 23:25 ` [PATCH v2 9/9] rpmsg: Turn name service into a stand alone driver Mathieu Poirier
2020-10-14 10:18 ` [PATCH v2 0/9] rpmsg: Make RPMSG name service modular Guennadi Liakhovetski
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201015194620.GE1450102@xps15 \
--to=mathieu.poirier@linaro.org \
--cc=arnaud.pouliquen@st.com \
--cc=bjorn.andersson@linaro.org \
--cc=guennadi.liakhovetski@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-remoteproc@vger.kernel.org \
--cc=ohad@wizery.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).