All of lore.kernel.org
 help / color / mirror / Atom feed
From: "An, Tedd" <tedd.an@intel.com>
To: "marcel@holtmann.org" <marcel@holtmann.org>,
	"tedd.an@linux.intel.com" <tedd.an@linux.intel.com>
Cc: "linux-bluetooth@vger.kernel.org" <linux-bluetooth@vger.kernel.org>
Subject: Re: [RFC PATCH V2] Bluetooth: vhci: Add support extended opcode for HCI Vendor packet
Date: Tue, 2 Nov 2021 21:44:09 +0000	[thread overview]
Message-ID: <c1ae1ce3b8090c7ca861200ac68dc73f43b08dae.camel@intel.com> (raw)
In-Reply-To: <F77C7BC5-4C6E-439F-8FEC-209E54F7BAAC@holtmann.org>

Hi Marcel,

On Tue, 2021-11-02 at 09:31 +0100, Marcel Holtmann wrote:
> Hi Tedd,
> 
> > Current implementation uses the Vendor packet type (0xff) with opcode
> > parameter. But there is no way to expand the opcode and no available bits
> > to use. Also it cannot be changed due to the backward compatibility
> > with older kernel.
> > 
> > THis patch adds new opcode(0x03) for HCI Vendor packet for VHCI for
> > extended device creation. This new opcode will not conflict with
> > existing legacy opcode because the legacy expects to set either bit 0 or
> > bit 1, but not both of bits.
> > 
> > It aslo requires new extra parameters for device type and flags to apply
> > to the VHCI device.
> > 
> > Signed-off-by: Tedd Ho-Jeong An <tedd.an@intel.com>
> > ---
> > drivers/bluetooth/hci_vhci.c | 65 ++++++++++++++++++++++++++++++++----
> > 1 file changed, 58 insertions(+), 7 deletions(-)
> > 
> > diff --git a/drivers/bluetooth/hci_vhci.c b/drivers/bluetooth/hci_vhci.c
> > index 49ac884d996e..5fccab136543 100644
> > --- a/drivers/bluetooth/hci_vhci.c
> > +++ b/drivers/bluetooth/hci_vhci.c
> > @@ -30,6 +30,16 @@
> > 
> > static bool amp;
> > 
> > +#define VHCI_EXT_OPCODE                                0x03
> > +struct vhci_ext_config {
> > +       __u8  dev_type;
> > +       __u32 flags;
> > +} __packed;
> > +
> > +#define VHCI_FLAG_QUIRK_RAW_DEVICE             0x01
> > +#define VHCI_FLAG_QUIRK_EXTERNAL_CONFIG                0x02
> > +#define VHCI_FLAG_QUIRKS_INVALID_BDADDR                0x04
> > +
> > struct vhci_data {
> >         struct hci_dev *hdev;
> > 
> > @@ -278,7 +288,8 @@ static int vhci_setup(struct hci_dev *hdev)
> >         return 0;
> > }
> > 
> > -static int __vhci_create_device(struct vhci_data *data, __u8 opcode)
> > +static int __vhci_create_device(struct vhci_data *data, __u8 opcode,
> > +                               struct vhci_ext_config *ext_config)
> > {
> >         struct hci_dev *hdev;
> >         struct sk_buff *skb;
> > @@ -287,8 +298,20 @@ static int __vhci_create_device(struct vhci_data *data, __u8 opcode)
> >         if (data->hdev)
> >                 return -EBADFD;
> > 
> > -       /* bits 0-1 are dev_type (Primary or AMP) */
> > -       dev_type = opcode & 0x03;
> > +       /* In case of legacy opcode, it doesn't allow to have 0x03 as an opcode,
> > +        * So, it is ok to assume that device is in the extended device
> > +        * creation mode when the opcode is 0x03. Also, it is required to have
> > +        * a ext_config and check it here.
> > +        */
> > +       if (ext_config && opcode != VHCI_EXT_OPCODE)
> > +               return -EINVAL;
> > +
> > +       if (opcode == VHCI_EXT_OPCODE)
> > +               dev_type = ext_config->dev_type;
> > +       else {
> > +               /* bits 0-1 are dev_type (Primary or AMP) */
> > +               dev_type = opcode & 0x03;
> > +       }
> > 
> >         if (dev_type != HCI_PRIMARY && dev_type != HCI_AMP)
> >                 return -EINVAL;
> > @@ -331,6 +354,16 @@ static int __vhci_create_device(struct vhci_data *data, __u8 opcode)
> >         if (opcode & 0x80)
> >                 set_bit(HCI_QUIRK_RAW_DEVICE, &hdev->quirks);
> > 
> > +       /* Flags for extended configuration */
> > +       if (ext_config && opcode == VHCI_EXT_OPCODE) {
> > +               if (ext_config->flags & VHCI_FLAG_QUIRK_EXTERNAL_CONFIG)
> > +                       set_bit(HCI_QUIRK_EXTERNAL_CONFIG, &hdev->quirks);
> > +               if (ext_config->flags & VHCI_FLAG_QUIRK_RAW_DEVICE)
> > +                       set_bit(HCI_QUIRK_RAW_DEVICE, &hdev->quirks);
> > +               if (ext_config->flags & VHCI_FLAG_QUIRKS_INVALID_BDADDR)
> > +                       set_bit(HCI_QUIRK_INVALID_BDADDR, &hdev->quirks);
> > +       }
> > +
> >         if (hci_register_dev(hdev) < 0) {
> >                 BT_ERR("Can't register HCI device");
> >                 hci_free_dev(hdev);
> > @@ -364,12 +397,13 @@ static int __vhci_create_device(struct vhci_data *data, __u8 opcode)
> >         return 0;
> > }
> > 
> > -static int vhci_create_device(struct vhci_data *data, __u8 opcode)
> > +static int vhci_create_device(struct vhci_data *data, __u8 opcode,
> > +                             struct vhci_ext_config *ext_config)
> > {
> >         int err;
> > 
> >         mutex_lock(&data->open_mutex);
> > -       err = __vhci_create_device(data, opcode);
> > +       err = __vhci_create_device(data, opcode, ext_config);
> >         mutex_unlock(&data->open_mutex);
> > 
> >         return err;
> > @@ -379,6 +413,7 @@ static inline ssize_t vhci_get_user(struct vhci_data *data,
> >                                     struct iov_iter *from)
> > {
> >         size_t len = iov_iter_count(from);
> > +       struct vhci_ext_config *ext_config;
> >         struct sk_buff *skb;
> >         __u8 pkt_type, opcode;
> >         int ret;
> > @@ -419,6 +454,21 @@ static inline ssize_t vhci_get_user(struct vhci_data *data,
> >                 opcode = *((__u8 *) skb->data);
> >                 skb_pull(skb, 1);
> > 
> > +               /* This opcode(0x03) is for extended device creation and it
> > +                * requires the extra parameters for extra configuration.
> > +                */
> > +               if (opcode == 0x03) {
> > +                       if (skb->len != sizeof(*ext_config)) {
> > +                               kfree_skb(skb);
> > +                               return -EINVAL;
> > +                       }
> > +
> > +                       ext_config = (void *) (skb->data);
> > +                       ret = vhci_create_device(data, opcode, ext_config);
> > +                       kfree_skb(skb);
> > +                       goto done;
> > +               }
> > +
> >                 if (skb->len > 0) {
> >                         kfree_skb(skb);
> >                         return -EINVAL;
> > @@ -426,7 +476,7 @@ static inline ssize_t vhci_get_user(struct vhci_data *data,
> > 
> >                 kfree_skb(skb);
> > 
> > -               ret = vhci_create_device(data, opcode);
> > +               ret = vhci_create_device(data, opcode, NULL);
> >                 break;
> > 
> >         default:
> > @@ -434,6 +484,7 @@ static inline ssize_t vhci_get_user(struct vhci_data *data,
> >                 return -EINVAL;
> >         }
> > 
> > +done:
> >         return (ret < 0) ? ret : len;
> > }
> > 
> > @@ -526,7 +577,7 @@ static void vhci_open_timeout(struct work_struct *work)
> >         struct vhci_data *data = container_of(work, struct vhci_data,
> >                                               open_timeout.work);
> > 
> > -       vhci_create_device(data, amp ? HCI_AMP : HCI_PRIMARY);
> > +       vhci_create_device(data, amp ? HCI_AMP : HCI_PRIMARY, NULL);
> > }
> 
> I think this is a bit convoluted in the end.
> 
> diff --git a/drivers/bluetooth/hci_vhci.c b/drivers/bluetooth/hci_vhci.c
> index 49ac884d996e..ce33ed63d021 100644
> --- a/drivers/bluetooth/hci_vhci.c
> +++ b/drivers/bluetooth/hci_vhci.c
> @@ -419,14 +419,22 @@ static inline ssize_t vhci_get_user(struct vhci_data *data,
>                 opcode = *((__u8 *) skb->data);
>                 skb_pull(skb, 1);
>  
> -               if (skb->len > 0) {
> -                       kfree_skb(skb);
> -                       return -EINVAL;
> +               /* The dev_type 3 is used as an escape opcode for extension
> +                * handling. If dev_type is set to 3 all other bits must be
> +                * set to zero.
> +                */
> +               if (opcode == 0x03) {
> +                       if (skb->len < 1)
> +                               ret = -EINVAL;
> +                       else
> +                               ret = vhci_create_extended_device(data, skb);
> +               } else {
> +                       if (skb->len > 0)
> +                               ret = -EINVAL;
> +                       else
> +                               ret = vhci_create_device(data, opcode);
>                 }
> -
>                 kfree_skb(skb);
> -
> -               ret = vhci_create_device(data, opcode);
>                 break;
>  
> I don’t fully like the nesting yet, but I would do it something like that.
> 
> Main point is that we keep the old way as it is and create a new clean path since otherwise the code
> becomes really hard to follow if you have to look at it in a few month.
Got it. It seems my first patch is more close to it, except new Vendor code. I will combine it with
your code above.


> I would also just include virtio_bt.h since that is actually an UAPI header and shared with
> userspace properly. I currently made extension struct just >= 1 and that might be good enough. We
> can check what the flags size is in virtio space. Otherwise we might just to a flexible flags
> array after 0x03 opcode.
Do you mean to include virtio_bt.h to share the data sturct between userspace and the kerenl?
On my ubuntu 20.04, virtio_bt.h is not available for some reason (on generic kernel)

> 
> Something that Bluetooth uses for EIR/AD flags data type. So you have {opcode}, {flag_len},
> {flags}[0..n], {bt_config} as your structure. That means the default is then 0x03, 0x00, 0x00
> packet to be sent. And if you want to enable AOSP, you send 0x03, 0x01, 0x04, 0x00.
I will use these method for the parameters. I only thought to support the quarks but it can even
expand to enable/disable MSFT EXT with opcode and no need to use debugfs to set it.

> 
> 
> Regards
> 
> Marcel
> 


      reply	other threads:[~2021-11-02 21:44 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-02  2:56 [RFC PATCH V2] Bluetooth: vhci: Add support extended opcode for HCI Vendor packet Tedd Ho-Jeong An
2021-11-02  3:01 ` An, Tedd
2021-11-02  8:31 ` Marcel Holtmann
2021-11-02 21:44   ` An, Tedd [this message]

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=c1ae1ce3b8090c7ca861200ac68dc73f43b08dae.camel@intel.com \
    --to=tedd.an@intel.com \
    --cc=linux-bluetooth@vger.kernel.org \
    --cc=marcel@holtmann.org \
    --cc=tedd.an@linux.intel.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.