linux-kernel - Re: [PATCH 1/4] Bluetooth: btusb: mediatek: use readx_poll

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 13 Sep 2022 09:47:47 +0200
From:   AngeloGioacchino Del Regno 
        <angelogioacchino.delregno@...labora.com>
To:     sean.wang@...iatek.com, marcel@...tmann.org,
        johan.hedberg@...il.com, luiz.dentz@...il.com
Cc:     Soul.Huang@...iatek.com, YN.Chen@...iatek.com,
        Leon.Yen@...iatek.com, Eric-SY.Chang@...iatek.com,
        Deren.Wu@...iatek.com, km.lin@...iatek.com,
        robin.chiu@...iatek.com, Eddie.Chen@...iatek.com,
        ch.yeh@...iatek.com, posh.sun@...iatek.com, ted.huang@...iatek.com,
        Stella.Chang@...iatek.com, Tom.Chou@...iatek.com,
        steve.lee@...iatek.com, jsiuda@...gle.com, frankgor@...gle.com,
        abhishekpandit@...gle.com, michaelfsun@...gle.com,
        abhishekpandit@...omium.org, mcchou@...omium.org,
        shawnku@...gle.com, linux-bluetooth@...r.kernel.org,
        linux-mediatek@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/4] Bluetooth: btusb: mediatek: use readx_poll_timeout
 instead of open coding

Il 13/09/22 00:18, sean.wang@...iatek.com ha scritto:
> From: Sean Wang <sean.wang@...iatek.com>
> 
> Use readx_poll_timeout instead of open coding to poll the hardware reset
> status until it is done.
> 
> Signed-off-by: Sean Wang <sean.wang@...iatek.com>

Hello Sean, thanks for the patch!
However, there's something to improve...

> ---
>   drivers/bluetooth/btusb.c | 32 ++++++++++++++++++--------------
>   1 file changed, 18 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
> index c3daba17de7f..4dc9cae3e937 100644
> --- a/drivers/bluetooth/btusb.c
> +++ b/drivers/bluetooth/btusb.c

..snip..

> @@ -2910,18 +2918,14 @@ static void btusb_mtk_cmd_timeout(struct hci_dev *hdev)
>   	btusb_mtk_uhw_reg_write(data, MTK_BT_SUBSYS_RST, 0);
>   	btusb_mtk_uhw_reg_read(data, MTK_BT_SUBSYS_RST, &val);
>   
> -	/* Poll the register until reset is completed */
> -	do {
> -		btusb_mtk_uhw_reg_read(data, MTK_BT_MISC, &val);
> -		if (val & MTK_BT_RST_DONE) {
> -			bt_dev_dbg(hdev, "Bluetooth Reset Successfully");
> -			break;
> -		}
> +	err = readx_poll_timeout(btusb_mtk_reset_done, hdev, val,
> +				 val & MTK_BT_RST_DONE,
> +				 100000, 1000000);

I agree with using readx_poll_timeout() instead of open coding the same, but
there's a catch: this macro uses usleep_range(), which is meant to be used
for sleeping less than ~20ms.

Even the kerneldoc at include/linux/iopoll.h advertises that:

  * @sleep_us: Maximum time to sleep between reads in us (0
  *            tight-loops).  Should be less than ~20ms since usleep_range
  *            is used (see Documentation/timers/timers-howto.rst).

So, if there's any reason for which you can't sleep for less than 100ms
per iteration, I'm afraid that you can't use readx_poll_timeout()...
...otherwise, please change sleep_us to 20000 and keep the timeout at 1 sec.

Regards,
Angelo

> +	if (err < 0)
> +		bt_dev_err(hdev, "Reset timeout");
>   
> -		bt_dev_dbg(hdev, "Polling Bluetooth Reset CR");
> -		retry++;
> -		msleep(MTK_BT_RESET_WAIT_MS);
> -	} while (retry < MTK_BT_RESET_NUM_TRIES);
> +	if (val & MTK_BT_RST_DONE)
> +		bt_dev_dbg(hdev, "Bluetooth Reset Successfully");
>   
>   	btusb_mtk_id_get(data, 0x70010200, &val);
>   	if (!val)