On Thu, 14 Mar 2024 at 01:35, Chris Lew <quic_clew@xxxxxxxxxxx> wrote:
>
>
>
> On 3/11/2024 8:34 AM, Dmitry Baryshkov wrote:
> > @@ -133,11 +133,13 @@ static int pdr_register_listener(struct pdr_handle *pdr,
> > req.enable = enable;
> > strscpy(req.service_path, pds->service_path, sizeof(req.service_path));
> >
> > + mutex_lock(&pdr->lock);
> > ret = qmi_send_request(&pdr->notifier_hdl, &pds->addr,
> > &txn, SERVREG_REGISTER_LISTENER_REQ,
> > SERVREG_REGISTER_LISTENER_REQ_LEN,
> > servreg_register_listener_req_ei,
> > &req);
> > + mutex_unlock(&pdr->lock);
> > if (ret < 0) {
> > qmi_txn_cancel(&txn);
> > return ret;
> >
>
> Hi Dmitry,
>
> What is the reason for taking the pdr lock here? The addr struct passed
> into qmi_send_request is from the pdr_service. I think this is different
> from the pdr_handle we are protecting in the other parts of the patch.
Indeed, we should be taking pdr->list_lock.
--
With best wishes
Dmitry
