Date:   Fri, 4 Dec 2020 11:25:29 +0800
From:   Archie Pusaka <apusaka@...gle.com>
To:     Marcel Holtmann <marcel@...tmann.org>
Cc:     linux-bluetooth <linux-bluetooth@...r.kernel.org>,
        CrosBT Upstreaming <chromeos-bluetooth-upstreaming@...omium.org>,
        Archie Pusaka <apusaka@...omium.org>,
        Miao-chen Chou <mcchou@...omium.org>,
        Yun-Hao Chung <howardchung@...gle.com>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Johan Hedberg <johan.hedberg@...il.com>,
        LKML <linux-kernel@...r.kernel.org>,
        "open list:NETWORKING [GENERAL]" <netdev@...r.kernel.org>
Subject: Re: [PATCH v1 1/5] Bluetooth: advmon offload MSFT add rssi support

Hi Marcel,

On Thu, 3 Dec 2020 at 22:03, Marcel Holtmann <marcel@...tmann.org> wrote:
>
> Hi Archie,
>
> > MSFT needs rssi parameter for monitoring advertisement packet,
> > therefore we should supply them from mgmt.
> >
> > Signed-off-by: Archie Pusaka <apusaka@...omium.org>
> > Reviewed-by: Miao-chen Chou <mcchou@...omium.org>
> > Reviewed-by: Yun-Hao Chung <howardchung@...gle.com>
>
> I don’t need any Reviewed-by if they are not catching an obvious user API breakage.
>
> > ---
> >
> > include/net/bluetooth/hci_core.h | 9 +++++++++
> > include/net/bluetooth/mgmt.h     | 9 +++++++++
> > net/bluetooth/mgmt.c             | 8 ++++++++
> > 3 files changed, 26 insertions(+)
> >
> > diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h
> > index 9873e1c8cd16..42d446417817 100644
> > --- a/include/net/bluetooth/hci_core.h
> > +++ b/include/net/bluetooth/hci_core.h
> > @@ -246,8 +246,17 @@ struct adv_pattern {
> >       __u8 value[HCI_MAX_AD_LENGTH];
> > };
> >
> > +struct adv_rssi_thresholds {
> > +     __s8 low_threshold;
> > +     __s8 high_threshold;
> > +     __u16 low_threshold_timeout;
> > +     __u16 high_threshold_timeout;
> > +     __u8 sampling_period;
> > +};
> > +
> > struct adv_monitor {
> >       struct list_head patterns;
> > +     struct adv_rssi_thresholds rssi;
> >       bool            active;
> >       __u16           handle;
> > };
> > diff --git a/include/net/bluetooth/mgmt.h b/include/net/bluetooth/mgmt.h
> > index d8367850e8cd..dc534837be0e 100644
> > --- a/include/net/bluetooth/mgmt.h
> > +++ b/include/net/bluetooth/mgmt.h
> > @@ -763,9 +763,18 @@ struct mgmt_adv_pattern {
> >       __u8 value[31];
> > } __packed;
> >
> > +struct mgmt_adv_rssi_thresholds {
> > +     __s8 high_threshold;
> > +     __le16 high_threshold_timeout;
> > +     __s8 low_threshold;
> > +     __le16 low_threshold_timeout;
> > +     __u8 sampling_period;
> > +} __packed;
> > +
> > #define MGMT_OP_ADD_ADV_PATTERNS_MONITOR      0x0052
> > struct mgmt_cp_add_adv_patterns_monitor {
> >       __u8 pattern_count;
> > +     struct mgmt_adv_rssi_thresholds rssi;
> >       struct mgmt_adv_pattern patterns[];
> > } __packed;
>
> This is something we can not do. It breaks an userspace facing API. Is the mgmt opcode 0x0052 in an already released kernel?

Yes, the opcode does exist in an already released kernel.

The DBus method which accesses this API is put behind the experimental
flag, therefore we expect they are flexible enough to support changes.
Previously, we already had a discussion in an email thread with the
title "Offload RSSI tracking to controller", and the outcome supports
this change.

Here is an excerpt of the discussion.
On Thu, 1 Oct 2020 at 05:58, Miao-chen Chou <mcchou@...gle.com> wrote:
>
> Hi Luiz,
>
> Yes, the RSSI is included as a part of the Adv monitor API, and the RSSI tracking is currently implemented (the patch series is still under review) in bluetoothd and used by bluetoothctl (submenu advmon). As mentioned, we are planning to offload the RSSI tracking to the controller as well, so there will be changes to the corresponding MGMT commands.
> Thanks for your quick feedback!
>
> Regards,
> Miao
>
> On Wed, Sep 30, 2020 at 2:00 PM Von Dentz, Luiz <luiz.von.dentz@...el.com> wrote:
>>
>> Hi Miao,
>>
>> I do recall seeing these at D-Bus level, or perhaps it was in use by bluetoothctl commands? Anyway since these are still experimental it should be fine to change them.
>> ________________________________
>> From: Miao-chen Chou <mcchou@...gle.com>
>> Sent: Wednesday, September 30, 2020 12:51 PM
>> To: Holtmann, Marcel <marcel.holtmann@...el.com>; Von Dentz, Luiz <luiz.von.dentz@...el.com>
>> Cc: Alain Michaud <alainmichaud@...gle.com>; Yun-hao Chung <howardchung@...gle.com>; Manish Mandlik <mmandlik@...gle.com>; Archie Pusaka <apusaka@...gle.com>
>> Subject: Offload RSSI tracking to controller.
>>
>> Hi Luiz and Marcel,
>>
>> Going forward to 2020 Q4, we will be working on offloading the content filtering to the controllers based on controll's support of MSFT HCI extension. In the meantime, we are planning to change the existing MGMT commands of Adv monitoring to allow the offloading of RSSI tracking shortly. Here is a snippet of potential changes.
>>
>> +struct mgmt_adv_rssi_thresholds {
>> +       __s8 high_rssi_threshold;
>> +       u16 high_rssi_threshold_timeout;
>> +       __s8 low_rssi_threshold;
>> +       u16 high_rssi_threshold_timeout;
>> +} __packed;
>>
>> struct mgmt_cp_add_adv_patterns_monitor {
>>         u8 pattern_count;
>> +        struct mgmt_adv_rssi_thresholds rssi_thresholds;
>>         struct mgmt_adv_pattern patterns[];
>> } __packed;
>>
>> Note that as suggested by you, the D-Bus Adv monitor API which accesses these MGMT commands is currently hidden behind the experimental flag, so they are still mutable. We'd like to hear your early feedback on changing the corresponding MGMT commands.
>>
>> Thanks,
>> Miao

Thanks,
Archie

Powered by blists - more mailing lists