From: Arseniy Krasnov <avkrasnov@salutedevices.com>
To: Luiz Augusto von Dentz <luiz.dentz@gmail.com>
Cc: Marcel Holtmann <marcel@holtmann.org>,
<linux-bluetooth@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
<oxffffaa@gmail.com>, <ddrokosov@salutedevices.com>
Subject: Re: [PATCH v1] Bluetooth: hci_uart: fix race during initialization
Date: Mon, 16 Dec 2024 18:06:36 +0300 [thread overview]
Message-ID: <609417f4-68c8-a382-0d95-04000c9df5cf@salutedevices.com> (raw)
In-Reply-To: <CABBYNZL6ZJkf=jQpdUyxYxZZwqRjHxWzDk4nCb74xdB78-33yg@mail.gmail.com>
On 16.12.2024 17:52, Luiz Augusto von Dentz wrote:
> Hi Arseniy,
>
> On Wed, Nov 27, 2024 at 7:38 AM Arseniy Krasnov
> <avkrasnov@salutedevices.com> wrote:
>>
>> 'hci_register_dev()' calls power up function, which is executed by
>> kworker - 'hci_power_on()'. This function does access to bluetooth chip
>> using callbacks from 'hci_ldisc.c', for example 'hci_uart_send_frame()'.
>> Now 'hci_uart_send_frame()' checks 'HCI_UART_PROTO_READY' bit set, and
>> if not - it fails. Problem is that 'HCI_UART_PROTO_READY' is set after
>> 'hci_register_dev()', and there is tiny chance that 'hci_power_on()' will
>> be executed before setting this bit. In that case HCI init logic fails.
>>
>> Patch adds one more bit in addition to 'HCI_UART_PROTO_READY' which
>> allows to perform power up logic without original bit set.
>>
>> Signed-off-by: Arseniy Krasnov <avkrasnov@salutedevices.com>
>> ---
>> drivers/bluetooth/hci_ldisc.c | 19 ++++++++++++++-----
>> drivers/bluetooth/hci_uart.h | 1 +
>> 2 files changed, 15 insertions(+), 5 deletions(-)
>>
>> diff --git a/drivers/bluetooth/hci_ldisc.c b/drivers/bluetooth/hci_ldisc.c
>> index 30192bb083549..2e20f00649bd7 100644
>> --- a/drivers/bluetooth/hci_ldisc.c
>> +++ b/drivers/bluetooth/hci_ldisc.c
>> @@ -102,7 +102,8 @@ static inline struct sk_buff *hci_uart_dequeue(struct hci_uart *hu)
>> if (!skb) {
>> percpu_down_read(&hu->proto_lock);
>>
>> - if (test_bit(HCI_UART_PROTO_READY, &hu->flags))
>> + if (test_bit(HCI_UART_PROTO_READY, &hu->flags) ||
>> + test_bit(HCI_UART_PROTO_INIT, &hu->flags))
>> skb = hu->proto->dequeue(hu);
>>
>> percpu_up_read(&hu->proto_lock);
>> @@ -124,7 +125,8 @@ int hci_uart_tx_wakeup(struct hci_uart *hu)
>> if (!percpu_down_read_trylock(&hu->proto_lock))
>> return 0;
>>
>> - if (!test_bit(HCI_UART_PROTO_READY, &hu->flags))
>> + if (!test_bit(HCI_UART_PROTO_READY, &hu->flags) &&
>> + !test_bit(HCI_UART_PROTO_INIT, &hu->flags))
>> goto no_schedule;
>>
>> set_bit(HCI_UART_TX_WAKEUP, &hu->tx_state);
>> @@ -278,7 +280,8 @@ static int hci_uart_send_frame(struct hci_dev *hdev, struct sk_buff *skb)
>>
>> percpu_down_read(&hu->proto_lock);
>>
>> - if (!test_bit(HCI_UART_PROTO_READY, &hu->flags)) {
>> + if (!test_bit(HCI_UART_PROTO_READY, &hu->flags) &&
>> + !test_bit(HCI_UART_PROTO_INIT, &hu->flags)) {
>> percpu_up_read(&hu->proto_lock);
>> return -EUNATCH;
>> }
>> @@ -582,7 +585,8 @@ static void hci_uart_tty_wakeup(struct tty_struct *tty)
>> if (tty != hu->tty)
>> return;
>>
>> - if (test_bit(HCI_UART_PROTO_READY, &hu->flags))
>> + if (test_bit(HCI_UART_PROTO_READY, &hu->flags) ||
>> + test_bit(HCI_UART_PROTO_INIT, &hu->flags))
>> hci_uart_tx_wakeup(hu);
>> }
>>
>> @@ -608,7 +612,8 @@ static void hci_uart_tty_receive(struct tty_struct *tty, const u8 *data,
>>
>> percpu_down_read(&hu->proto_lock);
>>
>> - if (!test_bit(HCI_UART_PROTO_READY, &hu->flags)) {
>> + if (!test_bit(HCI_UART_PROTO_READY, &hu->flags) &&
>> + !test_bit(HCI_UART_PROTO_INIT, &hu->flags)) {
>> percpu_up_read(&hu->proto_lock);
>> return;
>> }
>> @@ -704,12 +709,16 @@ static int hci_uart_set_proto(struct hci_uart *hu, int id)
>>
>> hu->proto = p;
>>
>> + set_bit(HCI_UART_PROTO_INIT, &hu->flags);
>> +
>> err = hci_uart_register_dev(hu);
>> if (err) {
>> return err;
>> }
>>
>> set_bit(HCI_UART_PROTO_READY, &hu->flags);
>> + clear_bit(HCI_UART_PROTO_INIT, &hu->flags);
>
> This should be quite obvious, why don't you just move the
> HCI_UART_PROTO_READY above hci_uart_register_dev?
>
Hi, this is simplest way, but I had doubts that this may be so obvious because
I don't have enough experience in bt logic.
I'll send v2 with it.
Thanks
>> return 0;
>> }
>>
>> diff --git a/drivers/bluetooth/hci_uart.h b/drivers/bluetooth/hci_uart.h
>> index 00bf7ae82c5b7..39f39704be41f 100644
>> --- a/drivers/bluetooth/hci_uart.h
>> +++ b/drivers/bluetooth/hci_uart.h
>> @@ -89,6 +89,7 @@ struct hci_uart {
>> #define HCI_UART_REGISTERED 1
>> #define HCI_UART_PROTO_READY 2
>> #define HCI_UART_NO_SUSPEND_NOTIFIER 3
>> +#define HCI_UART_PROTO_INIT 4
>>
>> /* TX states */
>> #define HCI_UART_SENDING 1
>> --
>> 2.30.1
>
>
>
next prev parent reply other threads:[~2024-12-16 15:06 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-11-27 12:38 Arseniy Krasnov
2024-12-16 11:58 ` Arseniy Krasnov
2024-12-16 14:52 ` Luiz Augusto von Dentz
2024-12-16 15:06 ` Arseniy Krasnov [this message]
2025-02-12 9:52 Arseniy Krasnov
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=609417f4-68c8-a382-0d95-04000c9df5cf@salutedevices.com \
--to=avkrasnov@salutedevices.com \
--cc=ddrokosov@salutedevices.com \
--cc=linux-bluetooth@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=luiz.dentz@gmail.com \
--cc=marcel@holtmann.org \
--cc=oxffffaa@gmail.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 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®