Dear guys,
I agree this patch.
BRs,
Max
-----Original Message-----
From: Kai-Heng Feng <kai.heng.feng@xxxxxxxxxxxxx>
Sent: Monday, November 2, 2020 9:33 PM
To: Max Chou <max.chou@xxxxxxxxxxx>; alex_lu <alex_lu@xxxxxxxxxxxxxx>
Cc: Marcel Holtmann <marcel@xxxxxxxxxxxx>; Johan Hedberg <johan.hedberg@xxxxxxxxx>; open list:BLUETOOTH DRIVERS <linux-bluetooth@xxxxxxxxxxxxxxx>; open list <linux-kernel@xxxxxxxxxxxxxxx>
Subject: Re: [PATCH v2] Bluetooth: btrtl: Ask 8821C to drop old firmware
> On Oct 26, 2020, at 16:28, Kai-Heng Feng <kai.heng.feng@xxxxxxxxxxxxx> wrote:
>
> Some platforms keep USB power even when they are powered off and in
> S5, this makes Realtek 8821C keep its firmware even after a cold boot,
> and make 8821C never load new firmware.
>
> So use vendor specific HCI command to ask 8821C drop its firmware
> after system shutdown.
>
> Newer firmware doesn't have this issue so we only use this trick for
> old 8821C firmware version.
>
> Suggested-by: Max Chou <max.chou@xxxxxxxxxxx>
> Signed-off-by: Kai-Heng Feng <kai.heng.feng@xxxxxxxxxxxxx>
Max and Alex,
Can you please ack or review the patch?
Kai-Heng
> ---
> v2:
> - Fix incorrect parAnthesis on le16_to_cpu.
> - Ensure firmware gets re-uploaded in initialization.
>
> drivers/bluetooth/btrtl.c | 46 +++++++++++++++++++++++++++++++++++++++
> 1 file changed, 46 insertions(+)
>
> diff --git a/drivers/bluetooth/btrtl.c b/drivers/bluetooth/btrtl.c
> index 3a9afc905f24..37e24bbb2eb4 100644
> --- a/drivers/bluetooth/btrtl.c
> +++ b/drivers/bluetooth/btrtl.c
> @@ -55,6 +55,7 @@ struct btrtl_device_info {
> int fw_len;
> u8 *cfg_data;
> int cfg_len;
> + bool drop_fw;
> };
>
> static const struct id_table ic_id_table[] = { @@ -563,6 +564,8 @@
> struct btrtl_device_info *btrtl_initialize(struct hci_dev *hdev,
> u16 hci_rev, lmp_subver;
> u8 hci_ver;
> int ret;
> + u16 opcode;
> + u8 cmd[2];
>
> btrtl_dev = kzalloc(sizeof(*btrtl_dev), GFP_KERNEL);
> if (!btrtl_dev) {
> @@ -584,6 +587,49 @@ struct btrtl_device_info *btrtl_initialize(struct hci_dev *hdev,
> hci_ver = resp->hci_ver;
> hci_rev = le16_to_cpu(resp->hci_rev);
> lmp_subver = le16_to_cpu(resp->lmp_subver);
> +
> + if (resp->hci_ver == 0x8 && le16_to_cpu(resp->hci_rev) == 0x826c &&
> + resp->lmp_ver == 0x8 && le16_to_cpu(resp->lmp_subver) == 0xa99e)
> + btrtl_dev->drop_fw = true;
> +
> + if (btrtl_dev->drop_fw) {
> + opcode = hci_opcode_pack(0x3f, 0x66);
> + cmd[0] = opcode & 0xff;
> + cmd[1] = opcode >> 8;
> +
> + skb = bt_skb_alloc(sizeof(cmd), GFP_KERNEL);
> + if (IS_ERR(skb))
> + goto out_free;
> +
> + skb_put_data(skb, cmd, sizeof(cmd));
> + hci_skb_pkt_type(skb) = HCI_COMMAND_PKT;
> +
> + hdev->send(hdev, skb);
> +
> + /* Ensure the above vendor command is sent to controller and
> + * process has done.
> + */
> + msleep(200);
> +
> + /* Read the local version again. Expect to have the vanilla
> + * version as cold boot.
> + */
> + skb = btrtl_read_local_version(hdev);
> + if (IS_ERR(skb)) {
> + ret = PTR_ERR(skb);
> + goto err_free;
> + }
> +
> + resp = (struct hci_rp_read_local_version *)skb->data;
> + rtl_dev_info(hdev, "examining hci_ver=%02x hci_rev=%04x lmp_ver=%02x lmp_subver=%04x",
> + resp->hci_ver, resp->hci_rev,
> + resp->lmp_ver, resp->lmp_subver);
> +
> + hci_ver = resp->hci_ver;
> + hci_rev = le16_to_cpu(resp->hci_rev);
> + lmp_subver = le16_to_cpu(resp->lmp_subver);
> + }
> +out_free:
> kfree_skb(skb);
>
> btrtl_dev->ic_info = btrtl_match_ic(lmp_subver, hci_rev, hci_ver,
> --
> 2.17.1
>
------Please consider the environment before printing this e-mail.
