mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Marcel Holtmann <marcel@holtmann.org>
To: Joseph Hwang <josephsih@chromium.org>
Cc: linux-bluetooth <linux-bluetooth@vger.kernel.org>,
	Luiz Augusto von Dentz <luiz.dentz@gmail.com>,
	pali@kernel.org, chromeos-bluetooth-upstreaming@chromium.org,
	josephsih@google.com, "David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>,
	Johan Hedberg <johan.hedberg@gmail.com>,
	linux-kernel@vger.kernel.org, netdev@vger.kernel.org
Subject: Re: [PATCH v6 1/3] Bluetooth: Add struct of reading AOSP vendor capabilities
Date: Mon, 25 Oct 2021 15:18:11 +0200	[thread overview]
Message-ID: <4E4ECD59-527D-4FDB-8EF8-21BC99FDE8D8@holtmann.org> (raw)
In-Reply-To: <20211021230356.v6.1.I139e71adfd3f00b88fe9edb63d013f9cd3e24506@changeid>

Hi Jospeh,

> This patch adds the struct of reading AOSP vendor capabilities.
> New capabilities are added incrementally. Note that the
> version_supported octets will be used to determine whether a
> capability has been defined for the version.
> 
> Signed-off-by: Joseph Hwang <josephsih@chromium.org>
> 
> ---
> 
> Changes in v6:
> - Add historical versions of struct aosp_rp_le_get_vendor_capabilities.
> - Perform the basic check about the struct length.
> - Through the version, bluetooth_quality_report_support can be checked.
> 
> Changes in v5:
> - This is a new patch.
> - Add struct aosp_rp_le_get_vendor_capabilities so that next patch
>  can determine whether a particular capability is supported or not.
> 
> include/net/bluetooth/hci_core.h |   1 +
> net/bluetooth/aosp.c             | 116 ++++++++++++++++++++++++++++++-
> 2 files changed, 116 insertions(+), 1 deletion(-)
> 
> diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h
> index dd8840e70e25..32b3774227f2 100644
> --- a/include/net/bluetooth/hci_core.h
> +++ b/include/net/bluetooth/hci_core.h
> @@ -603,6 +603,7 @@ struct hci_dev {
> 
> #if IS_ENABLED(CONFIG_BT_AOSPEXT)
> 	bool			aosp_capable;
> +	bool			aosp_quality_report;
> #endif
> 
> 	int (*open)(struct hci_dev *hdev);
> diff --git a/net/bluetooth/aosp.c b/net/bluetooth/aosp.c
> index a1b7762335a5..64684b2bf79b 100644
> --- a/net/bluetooth/aosp.c
> +++ b/net/bluetooth/aosp.c
> @@ -8,9 +8,53 @@
> 
> #include "aosp.h"
> 
> +/* Command complete parameters of LE_Get_Vendor_Capabilities_Command
> + * The parameters grow over time. The first version that declares the
> + * version_supported field is v0.95. Refer to
> + * https://cs.android.com/android/platform/superproject/+/master:system/
> + *         bt/gd/hci/controller.cc;l=452?q=le_get_vendor_capabilities_handler
> + */
> +
> +/* the base capabilities struct with the version_supported field */
> +struct aosp_rp_le_get_vendor_capa_v95 {
> +	__u8	status;
> +	__u8	max_advt_instances;
> +	__u8	offloaded_resolution_of_private_address;
> +	__u16	total_scan_results_storage;
> +	__u8	max_irk_list_sz;
> +	__u8	filtering_support;
> +	__u8	max_filter;
> +	__u8	activity_energy_info_support;
> +	__u16	version_supported;
> +	__u16	total_num_of_advt_tracked;
> +	__u8	extended_scan_support;
> +	__u8	debug_logging_supported;
> +} __packed;
> +
> +struct aosp_rp_le_get_vendor_capa_v96 {
> +	struct aosp_rp_le_get_vendor_capa_v95 v95;
> +	/* v96 */
> +	__u8	le_address_generation_offloading_support;
> +} __packed;
> +
> +struct aosp_rp_le_get_vendor_capa_v98 {
> +	struct aosp_rp_le_get_vendor_capa_v96 v96;
> +	/* v98 */
> +	__u32	a2dp_source_offload_capability_mask;
> +	__u8	bluetooth_quality_report_support;
> +} __packed;
> +
> +struct aosp_rp_le_get_vendor_capa_v100 {
> +	struct aosp_rp_le_get_vendor_capa_v98 v98;
> +	/* v100 */
> +	__u32	dynamic_audio_buffer_support;
> +} __packed;
> +

while this is an interesting idea to document the different versions, I don’t thing you need to overboard like this. Just use the full struct and make sure you check the version_supported to ensure you are not reading beyond the bounds of your result.

Meaning you can have a struct pointer point to smaller memory, just make sure you never access it if it is not there. So I would just add defines for the different versions and their length of the struct.

Oh, I might have forgotten that, you need to use __le16 and __le32 to indicate the endianness of the data on the wire.

> void aosp_do_open(struct hci_dev *hdev)
> {
> 	struct sk_buff *skb;
> +	struct aosp_rp_le_get_vendor_capa_v95 *base_rp;
> +	u16 version_supported;
> 
> 	if (!hdev->aosp_capable)
> 		return;
> @@ -20,9 +64,79 @@ void aosp_do_open(struct hci_dev *hdev)
> 	/* LE Get Vendor Capabilities Command */
> 	skb = __hci_cmd_sync(hdev, hci_opcode_pack(0x3f, 0x153), 0, NULL,
> 			     HCI_CMD_TIMEOUT);
> -	if (IS_ERR(skb))
> +	if (IS_ERR(skb)) {
> +		bt_dev_warn(hdev, "AOSP get vendor capabilities (%ld)",
> +			    PTR_ERR(skb));

This is actually an error. If the driver indicates support for it, this better succeed. If not complain loudly.

> 		return;
> +	}
> +
> +	bt_dev_dbg(hdev, "aosp le vendor capabilities length %d", skb->len);

Skip this one.

Add a basic length check that you know you can get to rp->version_supported field.

> +
> +	base_rp = (struct aosp_rp_le_get_vendor_capa_v95 *)skb->data;
> +
> +	if (base_rp->status) {
> +		bt_dev_err(hdev, "AOSP LE Get Vendor Capabilities status %d",
> +			   base_rp->status);
> +		goto done;
> +	}

Actually the status is already evaluated via the __hci_cmd_sync command. No need to repeat it here.

> +
> +	version_supported = le16_to_cpu(base_rp->version_supported);
> +	bt_dev_info(hdev, "AOSP version %u", version_supported);
> +
> +	/* Do not support very old versions. */
> +	if (version_supported < 95) {
> +		bt_dev_err(hdev, "capabilities version %u too old",
> +			   version_supported);
> +		goto done;
> +	}

This is not an error. Just print a warning here. And “AOSP capabilities ..” please.

> +
> +	if (version_supported >= 95) {
> +		struct aosp_rp_le_get_vendor_capa_v95 *rp;
> +
> +		rp = (struct aosp_rp_le_get_vendor_capa_v95 *)skb->data;
> +		if (skb->len < sizeof(*rp))
> +			goto length_error;
> +	}
> +
> +	if (version_supported >= 96) {
> +		struct aosp_rp_le_get_vendor_capa_v96 *rp;
> +
> +		rp = (struct aosp_rp_le_get_vendor_capa_v96 *)skb->data;
> +		if (skb->len < sizeof(*rp))
> +			goto length_error;
> +	}

Since we don’t use any data out of these two above, skip it. If the version is less than < 0.98 we are just ignoring it. You can print a warning that "AOSP quality report is not supported”.


> +
> +	if (version_supported >= 98) {
> +		struct aosp_rp_le_get_vendor_capa_v98 *rp;
> +
> +		rp = (struct aosp_rp_le_get_vendor_capa_v98 *)skb->data;
> +		if (skb->len < sizeof(*rp))
> +			goto length_error;
> +
> +		/* The bluetooth_quality_report_support is defined at version v0.98.
> +		 * Refer to https://cs.android.com/android/platform/superproject/+/
> +		 *                  master:system/bt/gd/hci/controller.cc;l=477
> +		 */
> +		if (rp->bluetooth_quality_report_support) {
> +			hdev->aosp_quality_report = true;
> +			bt_dev_info(hdev, "bluetooth quality report is supported");
> +		}
> +	}
> +
> +	if (version_supported >= 100) {
> +		struct aosp_rp_le_get_vendor_capa_v100 *rp;
> +
> +		rp = (struct aosp_rp_le_get_vendor_capa_v100 *)skb->data;
> +		if (skb->len < sizeof(*rp))
> +			goto length_error;
> +	}

Skip that one as well. We only care about quality report support.

> +
> +	goto done;
> +
> +length_error:
> +	bt_dev_err(hdev, "AOSP capabilities length %d too short", skb->len);
> 
> +done:
> 	kfree_skb(skb);
> }

Regards

Marcel


      parent reply	other threads:[~2021-10-25 13:18 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-10-21 15:04 Joseph Hwang
2021-10-21 15:04 ` [PATCH v6 2/3] Bluetooth: aosp: Support AOSP Bluetooth Quality Report Joseph Hwang
2021-10-25 13:24   ` Marcel Holtmann
2021-10-21 15:04 ` [PATCH v6 3/3] Bluetooth: btusb: enable Mediatek to support AOSP extension Joseph Hwang
2021-10-25 13:26   ` Marcel Holtmann
2021-10-29  8:08     ` Joseph Hwang
2021-10-29  8:21       ` Marcel Holtmann
2021-10-29 14:12         ` Joseph Hwang
2021-10-25 13:18 ` Marcel Holtmann [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=4E4ECD59-527D-4FDB-8EF8-21BC99FDE8D8@holtmann.org \
    --to=marcel@holtmann.org \
    --cc=chromeos-bluetooth-upstreaming@chromium.org \
    --cc=davem@davemloft.net \
    --cc=johan.hedberg@gmail.com \
    --cc=josephsih@chromium.org \
    --cc=josephsih@google.com \
    --cc=kuba@kernel.org \
    --cc=linux-bluetooth@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=luiz.dentz@gmail.com \
    --cc=netdev@vger.kernel.org \
    --cc=pali@kernel.org \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®