Date:   Tue, 7 Dec 2021 10:48:06 +0100
From:   Loic Poulain <loic.poulain@...aro.org>
To:     Ricardo Martinez <ricardo.martinez@...ux.intel.com>
Cc:     netdev@...r.kernel.org, linux-wireless@...r.kernel.org,
        kuba@...nel.org, davem@...emloft.net, johannes@...solutions.net,
        ryazanov.s.a@...il.com, m.chetan.kumar@...el.com,
        chandrashekar.devegowda@...el.com, linuxwwan@...el.com,
        chiranjeevi.rapolu@...ux.intel.com, haijun.liu@...iatek.com,
        amir.hanania@...el.com, andriy.shevchenko@...ux.intel.com,
        dinesh.sharma@...el.com, eliot.lee@...el.com,
        mika.westerberg@...ux.intel.com, moises.veleta@...el.com,
        pierre-louis.bossart@...el.com, muralidharan.sethuraman@...el.com,
        Soumya.Prakash.Mishra@...el.com, sreehari.kancharla@...el.com,
        suresh.nagaraj@...el.com
Subject: Re: [PATCH net-next v3 05/12] net: wwan: t7xx: Add AT and MBIM WWAN ports

Hi Ricardo,

Overall, it looks good, but:

On Tue, 7 Dec 2021 at 03:48, Ricardo Martinez
<ricardo.martinez@...ux.intel.com> wrote:
>
> From: Chandrashekar Devegowda <chandrashekar.devegowda@...el.com>
>
> Adds AT and MBIM ports to the port proxy infrastructure.
> The initialization method is responsible for creating the corresponding
> ports using the WWAN framework infrastructure. The implemented WWAN port
> operations are start, stop, and TX.
>
> Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@...el.com>
> Co-developed-by: Ricardo Martinez <ricardo.martinez@...ux.intel.com>
> Signed-off-by: Ricardo Martinez <ricardo.martinez@...ux.intel.com>
> ---
>  drivers/net/wwan/t7xx/Makefile          |   1 +
>  drivers/net/wwan/t7xx/t7xx_port_proxy.c |  24 +++
>  drivers/net/wwan/t7xx/t7xx_port_proxy.h |   1 +
>  drivers/net/wwan/t7xx/t7xx_port_wwan.c  | 258 ++++++++++++++++++++++++
>  4 files changed, 284 insertions(+)
>  create mode 100644 drivers/net/wwan/t7xx/t7xx_port_wwan.c
>
[...]
> +static int t7xx_port_ctrl_tx(struct wwan_port *port, struct sk_buff *skb)
> +{
> +       struct t7xx_port *port_private = wwan_port_get_drvdata(port);
> +       size_t actual_count = 0, alloc_size = 0, txq_mtu = 0;
> +       struct t7xx_port_static *port_static;
> +       int i, multi_packet = 1, ret = 0;
> +       struct sk_buff *skb_ccci = NULL;
> +       struct t7xx_fsm_ctl *ctl;
> +       enum md_state md_state;
> +       unsigned int count;
> +       bool port_multi;
> +
> +       count = skb->len;
> +       if (!count)
> +               return -EINVAL;
> +
> +       port_static = port_private->port_static;
> +       ctl = port_private->t7xx_dev->md->fsm_ctl;
> +       md_state = t7xx_fsm_get_md_state(ctl);
> +       if (md_state == MD_STATE_WAITING_FOR_HS1 || md_state == MD_STATE_WAITING_FOR_HS2) {
> +               dev_warn(port_private->dev, "Cannot write to %s port when md_state=%d\n",
> +                        port_static->name, md_state);
> +               return -ENODEV;
> +       }
> +
> +       txq_mtu = CLDMA_TXQ_MTU;
> +
> +       if (port_private->flags & PORT_F_USER_HEADER) {
> +               if (port_private->flags & PORT_F_USER_HEADER && count > txq_mtu) {
> +                       dev_err(port_private->dev, "Packet %u larger than MTU on %s port\n",
> +                               count, port_static->name);
> +                       return -ENOMEM;
> +               }
> +
> +               alloc_size = min_t(size_t, txq_mtu, count);
> +               actual_count = alloc_size;
> +       } else {
> +               alloc_size = min_t(size_t, txq_mtu, count + CCCI_H_ELEN);
> +               actual_count = alloc_size - CCCI_H_ELEN;
> +               port_multi = t7xx_port_wwan_multipkt_capable(port_static);
> +               if ((count + CCCI_H_ELEN > txq_mtu) && port_multi)
> +                       multi_packet = DIV_ROUND_UP(count, txq_mtu - CCCI_H_ELEN);
> +       }
> +
> +       for (i = 0; i < multi_packet; i++) {
> +               struct ccci_header *ccci_h = NULL;
> +
> +               if (multi_packet > 1 && multi_packet == i + 1) {
> +                       actual_count = count % (txq_mtu - CCCI_H_ELEN);
> +                       alloc_size = actual_count + CCCI_H_ELEN;
> +               }
> +
> +               skb_ccci = __dev_alloc_skb(alloc_size, GFP_KERNEL);
> +               if (!skb_ccci)
> +                       return -ENOMEM;
> +
> +               ccci_h = skb_put(skb_ccci, CCCI_H_LEN);
> +               ccci_h->packet_header = 0;
> +               ccci_h->packet_len = cpu_to_le32(actual_count + CCCI_H_LEN);
> +               ccci_h->status &= cpu_to_le32(~HDR_FLD_CHN);
> +               ccci_h->status |= cpu_to_le32(FIELD_PREP(HDR_FLD_CHN, port_static->tx_ch));
> +               ccci_h->ex_msg = 0;
> +
> +               memcpy(skb_put(skb_ccci, actual_count), skb->data + i * (txq_mtu - CCCI_H_ELEN),
> +                      actual_count);
> +
> +               t7xx_port_proxy_set_seq_num(port_private, ccci_h);
> +
> +               ret = t7xx_port_send_skb_to_md(port_private, skb_ccci, true);
> +               if (ret)
> +                       goto err_free_skb;
> +
> +               port_private->seq_nums[MTK_TX]++;
> +
> +               if (multi_packet == 1)
> +                       return actual_count;
> +               else if (multi_packet == i + 1)
> +                       return count;

wwan port tx ops is supposed to return 0 on success, and release the
processed skb. In your case it works because wwan core does:

ret = wwan_port_op_tx(port, skb, !!(filp->f_flags & O_NONBLOCK));
if (ret) {
        kfree_skb(skb);
        return ret;
}

So returning a positive value here (count) makes the core enter the
error case, release the skb for you and return your count value to
write fops. Eventually, you land on your feet, but it's not what was
expected from tx ops. You should simply return 0 on success and
release the skb. Alternatively, if you think partial TX is something
we need, it should be somewhat documented in the wwan_port_fops_write
function or tx_ops definition.

Regards,
Loic

Powered by blists - more mailing lists