Message-ID: <c5395ebf-b0ea-4be8-b0c0-6a51d4c98e09@amd.com>
Date: Wed, 3 Dec 2025 12:12:46 -0600
From: Tanmay Shah <tanmay.shah@....com>
To: Zhongqiu Han <zhongqiu.han@....qualcomm.com>, <andersson@...nel.org>,
	<mathieu.poirier@...aro.org>, <mst@...hat.com>, <jasowang@...hat.com>,
	<xuanzhuo@...ux.alibaba.com>, <eperezma@...hat.com>
CC: <linux-kernel@...r.kernel.org>, <linux-remoteproc@...r.kernel.org>,
	<virtualization@...ts.linux.dev>, <xiaoxiang@...omi.com>, Xiang Xiao
	<xiaoxiang781216@...il.com>, <arnaud.pouliquen@...s.st.com>
Subject: Re: [RFC PATCH 2/2] rpmsg: virtio_rpmsg_bus: get buffer size from
 config space

Hello,

Thanks for your reviews. Please find the response below.

On 11/22/25 6:05 AM, Zhongqiu Han wrote:
> On 11/15/2025 2:46 AM, Tanmay Shah wrote:
>> From: Xiang Xiao <xiaoxiang781216@...il.com>
>>
>> 512 bytes isn't always suitable for all case, let firmware
>> maker decide the best value from resource table.
>> enable by VIRTIO_RPMSG_F_BUFSZ feature bit.
>>
>> Signed-off-by: Xiang Xiao <xiaoxiang@...omi.com>
>> Signed-off-by: Tanmay Shah <tanmay.shah@....com>
>> ---
>>   drivers/rpmsg/virtio_rpmsg_bus.c | 68 +++++++++++++++++++++++---------
>>   include/linux/virtio_rpmsg.h     | 24 +++++++++++
>>   2 files changed, 74 insertions(+), 18 deletions(-)
>>   create mode 100644 include/linux/virtio_rpmsg.h
>>
>> diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c b/drivers/rpmsg/ 
>> virtio_rpmsg_bus.c
>> index cc26dfcc3e29..03dd5535880a 100644
>> --- a/drivers/rpmsg/virtio_rpmsg_bus.c
>> +++ b/drivers/rpmsg/virtio_rpmsg_bus.c
>> @@ -25,6 +25,7 @@
>>   #include <linux/sched.h>
>>   #include <linux/virtio.h>
>>   #include <linux/virtio_ids.h>
>> +#include <linux/virtio_rpmsg.h>
>>   #include <linux/virtio_config.h>
>>   #include <linux/wait.h>
>> @@ -39,7 +40,8 @@
>>    * @sbufs:    kernel address of tx buffers
>>    * @num_rbufs:    total number of buffers for rx
>>    * @num_sbufs:    total number of buffers for tx
>> - * @buf_size:    size of one rx or tx buffer
>> + * @rbuf_size:    size of one rx buffer
>> + * @sbuf_size:    size of one tx buffer
>>    * @last_sbuf:    index of last tx buffer used
>>    * @bufs_dma:    dma base addr of the buffers
>>    * @tx_lock:    protects svq, sbufs and sleepers, to allow 
>> concurrent senders.
>> @@ -60,7 +62,8 @@ struct virtproc_info {
>>       void *rbufs, *sbufs;
>>       unsigned int num_rbufs;
>>       unsigned int num_sbufs;
>> -    unsigned int buf_size;
>> +    unsigned int rbuf_size;
>> +    unsigned int sbuf_size;
>>       int last_sbuf;
>>       dma_addr_t bufs_dma;
>>       struct mutex tx_lock;
>> @@ -70,9 +73,6 @@ struct virtproc_info {
>>       atomic_t sleepers;
>>   };
>> -/* The feature bitmap for virtio rpmsg */
>> -#define VIRTIO_RPMSG_F_NS    0 /* RP supports name service 
>> notifications */
>> -
>>   /**
>>    * struct rpmsg_hdr - common header for all rpmsg messages
>>    * @src: source address
>> @@ -130,7 +130,7 @@ struct virtio_rpmsg_channel {
>>    * processor.
>>    */
>>   #define MAX_RPMSG_NUM_BUFS    (256)
>> -#define MAX_RPMSG_BUF_SIZE    (512)
>> +#define DEFAULT_RPMSG_BUF_SIZE    (512)
>>   /*
>>    * Local addresses are dynamically allocated on-demand.
>> @@ -443,7 +443,7 @@ static void *get_a_tx_buf(struct virtproc_info *vrp)
>>       /* either pick the next unused tx buffer */
>>       if (vrp->last_sbuf < vrp->num_sbufs)
>> -        ret = vrp->sbufs + vrp->buf_size * vrp->last_sbuf++;
>> +        ret = vrp->sbufs + vrp->sbuf_size * vrp->last_sbuf++;
>>       /* or recycle a used one */
>>       else
>>           ret = virtqueue_get_buf(vrp->svq, &len);
>> @@ -569,7 +569,7 @@ static int rpmsg_send_offchannel_raw(struct 
>> rpmsg_device *rpdev,
>>        * messaging), or to improve the buffer allocator, to support
>>        * variable-length buffer sizes.
>>        */
>> -    if (len > vrp->buf_size - sizeof(struct rpmsg_hdr)) {
>> +    if (len > vrp->sbuf_size - sizeof(struct rpmsg_hdr)) {
>>           dev_err(dev, "message is too big (%d)\n", len);
>>           return -EMSGSIZE;
>>       }
>> @@ -680,7 +680,7 @@ static ssize_t virtio_rpmsg_get_mtu(struct 
>> rpmsg_endpoint *ept)
>>       struct rpmsg_device *rpdev = ept->rpdev;
>>       struct virtio_rpmsg_channel *vch = to_virtio_rpmsg_channel(rpdev);
>> -    return vch->vrp->buf_size - sizeof(struct rpmsg_hdr);
>> +    return vch->vrp->sbuf_size - sizeof(struct rpmsg_hdr);
>>   }
>>   static int rpmsg_recv_single(struct virtproc_info *vrp, struct 
>> device *dev,
>> @@ -706,7 +706,7 @@ static int rpmsg_recv_single(struct virtproc_info 
>> *vrp, struct device *dev,
>>        * We currently use fixed-sized buffers, so trivially sanitize
>>        * the reported payload length.
>>        */
>> -    if (len > vrp->buf_size ||
>> +    if (len > vrp->rbuf_size ||
>>           msg_len > (len - sizeof(struct rpmsg_hdr))) {
>>           dev_warn(dev, "inbound msg too big: (%d, %d)\n", len, msg_len);
>>           return -EINVAL;
>> @@ -739,7 +739,7 @@ static int rpmsg_recv_single(struct virtproc_info 
>> *vrp, struct device *dev,
>>           dev_warn_ratelimited(dev, "msg received with no recipient\n");
>>       /* publish the real size of the buffer */
>> -    rpmsg_sg_init(&sg, msg, vrp->buf_size);
>> +    rpmsg_sg_init(&sg, msg, vrp->rbuf_size);
>>       /* add the buffer back to the remote processor's virtqueue */
>>       err = virtqueue_add_inbuf(vrp->rvq, &sg, 1, msg, GFP_KERNEL);
>> @@ -888,9 +888,39 @@ static int rpmsg_probe(struct virtio_device *vdev)
>>       else
>>           vrp->num_sbufs = MAX_RPMSG_NUM_BUFS;
>> -    vrp->buf_size = MAX_RPMSG_BUF_SIZE;
>> +    /*
>> +     * If VIRTIO_RPMSG_F_BUFSZ feature is supported, then configure buf
>> +     * size from virtio device config space from the resource table.
>> +     * If the feature is not supported, then assign default buf size.
>> +     */
>> +    if (virtio_has_feature(vdev, VIRTIO_RPMSG_F_BUFSZ)) {
>> +        /* note: virtio_rpmsg_config is defined from remote view */
>> +        virtio_cread(vdev, struct virtio_rpmsg_config,
>> +                 txbuf_size, &vrp->rbuf_size);
>> +        virtio_cread(vdev, struct virtio_rpmsg_config,
>> +                 rxbuf_size, &vrp->sbuf_size);
>> +
>> +        /* The buffers must hold rpmsg header atleast */
>> +        if (vrp->rbuf_size < sizeof(struct rpmsg_hdr) ||
>> +            vrp->sbuf_size < sizeof(struct rpmsg_hdr)) {
> 
> 
> Hello Tanmay,
> 
> May I know if the omission of = here is to accommodate the ping/pong/ack
> scenarios? mtu will 0
> 
> 

Yes. At minimum RPMsg header is needed to ping the correct endpoint. We 
don't need to have any payload attached to the packet. MTU will be 
sizeof rpmsg_hdr I think.

>> +            dev_err(&vdev->dev,
>> +                "vdev config: rx buf sz = %d, tx buf sz = %d\n",
>> +                vrp->rbuf_size, vrp->sbuf_size);
>> +            err = -EINVAL;
>> +            goto vqs_del;
>> +        }
>> +
>> +        dev_dbg(&vdev->dev,
>> +            "vdev config: rx buf sz = 0x%x, tx buf sz = 0x%x\n",
>> +            vrp->rbuf_size, vrp->sbuf_size);
>> +    } else {
>> +        vrp->rbuf_size = DEFAULT_RPMSG_BUF_SIZE;
>> +        vrp->sbuf_size = DEFAULT_RPMSG_BUF_SIZE;
>> +    }
>> -    total_buf_space = (vrp->num_rbufs + vrp->num_sbufs) * vrp->buf_size;
>> +    total_buf_space = (vrp->num_rbufs * vrp->rbuf_size) +
>> +              (vrp->num_sbufs * vrp->sbuf_size);
>> +    total_buf_space = ALIGN(total_buf_space, PAGE_SIZE);
>>       /* allocate coherent memory for the buffers */
>>       bufs_va = dma_alloc_coherent(vdev->dev.parent,
>> @@ -908,14 +938,14 @@ static int rpmsg_probe(struct virtio_device *vdev)
>>       vrp->rbufs = bufs_va;
>>       /* and second part is dedicated for TX */
>> -    vrp->sbufs = bufs_va + vrp->num_rbufs * vrp->buf_size;
>> +    vrp->sbufs = bufs_va + (vrp->num_rbufs * vrp->rbuf_size);
>>       /* set up the receive buffers */
>>       for (i = 0; i < vrp->num_rbufs; i++) {
>>           struct scatterlist sg;
>> -        void *cpu_addr = vrp->rbufs + i * vrp->buf_size;
>> +        void *cpu_addr = vrp->rbufs + i * vrp->rbuf_size;
>> -        rpmsg_sg_init(&sg, cpu_addr, vrp->buf_size);
>> +        rpmsg_sg_init(&sg, cpu_addr, vrp->rbuf_size);
>>           err = virtqueue_add_inbuf(vrp->rvq, &sg, 1, cpu_addr,
>>                         GFP_KERNEL);
>> @@ -1001,8 +1031,8 @@ static int rpmsg_remove_device(struct device 
>> *dev, void *data)
>>   static void rpmsg_remove(struct virtio_device *vdev)
>>   {
>>       struct virtproc_info *vrp = vdev->priv;
>> -    unsigned int num_bufs = vrp->num_rbufs + vrp->num_sbufs;
>> -    size_t total_buf_space = num_bufs * vrp->buf_size;
>> +    size_t total_buf_space = (vrp->num_rbufs * vrp->rbuf_size) +
>> +                 (vrp->num_sbufs * vrp->sbuf_size);
>>       int ret;
>>       virtio_reset_device(vdev);
>> @@ -1015,6 +1045,7 @@ static void rpmsg_remove(struct virtio_device 
>> *vdev)
>>       vdev->config->del_vqs(vrp->vdev);
>> +    total_buf_space = ALIGN(total_buf_space, PAGE_SIZE);
>>       dma_free_coherent(vdev->dev.parent, total_buf_space,
>>                 vrp->rbufs, vrp->bufs_dma);
>> @@ -1028,6 +1059,7 @@ static struct virtio_device_id id_table[] = {
>>   static unsigned int features[] = {
>>       VIRTIO_RPMSG_F_NS,
>> +    VIRTIO_RPMSG_F_BUFSZ,
>>   };
>>   static struct virtio_driver virtio_ipc_driver = {
>> diff --git a/include/linux/virtio_rpmsg.h b/include/linux/virtio_rpmsg.h
>> new file mode 100644
>> index 000000000000..6406bc505383
>> --- /dev/null
>> +++ b/include/linux/virtio_rpmsg.h
>> @@ -0,0 +1,24 @@
>> +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
> 
> 
> Echo Arnaud's comments. If it is intended for UAPI, please keep it in
> include/uapi/linux
> 
> 

It's not intended for UAPI. I need to fix the license. I will check 
other virtio headers in the same directory and fix the license accordingly.

>> +/*
>> + * Copyright (C) Pinecone Inc. 2019
>> + * Copyright (C) Xiang Xiao <xiaoxiang@...econe.net>
>> + */
>> +
>> +#ifndef _LINUX_VIRTIO_RPMSG_H
>> +#define _LINUX_VIRTIO_RPMSG_H
>> +
>> +#include <linux/types.h>
>> +
>> +/* The feature bitmap for virtio rpmsg */
>> +#define VIRTIO_RPMSG_F_NS    0 /* RP supports name service 
>> notifications */
>> +#define VIRTIO_RPMSG_F_BUFSZ    2 /* RP get buffer size from config 
>> space */
> 
> May I know why skip bit 1?
> 
> 

Thanks, that's a good question. I keept id 2 unmodified from the 
original series. I don't know why ID 2 was chosen in the original 
series. I will have to discuss this with the linux remoteproc/rpmsg 
maintainers and choose the correct ID.

I don't see any problem choosing ID 1, but for some reason if ID 1 was 
assigned and deprecated (I don't think that is the case) then only we 
should use ID 2.


Arnaud, Mathieu, Bjorn any input here?

>> +
>> +struct virtio_rpmsg_config {
>> +    /* The tx/rx individual buffer size(if VIRTIO_RPMSG_F_BUFSZ) */
>> +    __u32 txbuf_size;
>> +    __u32 rxbuf_size;
>> +    __u32 reserved[14]; /* Reserve for the future use */
> 
> Should we use __virtio32 instead of __u32 to avoid endianness issues?
> 
> 

Sure, if that is the standard in other virtio headers I will modify it.

Thanks,
Tanmay

>> +    /* Put the customize config here */
>> +} __attribute__((packed));
>> +
>> +#endif /* _LINUX_VIRTIO_RPMSG_H */
> 
> 

Powered by blists - more mailing lists