linux-kernel - Re: [PATCH v5] tty: serial: don't do termios for BTIF

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] [day] [month] [year] [list]

Message-ID: <20200708145814.GN3703480@smile.fi.intel.com>
Date:   Wed, 8 Jul 2020 17:58:14 +0300
From:   Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To:     sean.wang@...iatek.com
Cc:     gregkh@...uxfoundation.org, jslaby@...e.com,
        mika.westerberg@...ux.intel.com, sr@...x.de, arnd@...db.de,
        matthias.bgg@...il.com, tthayer@...nsource.altera.com,
        linux-mediatek@...ts.infradead.org, linux-serial@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        Steven Liu <steven.liu@...iatek.com>,
        Ryder Lee <ryder.lee@...iatek.com>
Subject: Re: [PATCH v5] tty: serial: don't do termios for BTIF

On Wed, Jul 08, 2020 at 07:11:22AM +0800, sean.wang@...iatek.com wrote:
> From: Sean Wang <sean.wang@...iatek.com>
> 
> Bluetooth Interface (BTIF) is designed dedicatedly for MediaTek SOC with
> BT in order to be instead of the UART interface between BT module and Host
> CPU, and not exported to user space to access.
> 
> As the UART design, BTIF will be an APB slave and can transmit or receive
> data by MCU access, but doesn't provide termios function like baudrate and
> flow control setup.
> 
> Even LCR on offset 0xC that is just a FAKELCR
> a. If FAKELCR[7] is equaled to 1, RBR(0x00), THR(0x00), IER(0x04)
>    will not be readable/writable.
> 
> b. If FAKELCR is equaled to 0xBF, RBR(0x00), THR(0x00), IER(0x04),
>    IIR(0x08), and LSR(0x14) will not be readable/writable.
> 
> So adding a new capability 'UART_CAP_NTIO' for the unusual unsupported
> case.
> 
> The bluetooth driver would use BTIF device as a serdev. So the termios
> still function would be called in kernelspace from ttyport_open in
> drivers/tty/serdev/serdev-ttyprt.c.

...
>  	up->port.quirks |= skip_txen_test ? UPQ_NO_TXEN_TEST : 0;
> +
> +	if (up->port.type == PORT_MTK_BTIF)
> +		up->port.quirks |= UPQ_IGNORE_TERMIOS;

I don't like to see this in core.

Can we have this inside 8250_port.c? Something like extending struct
serial8250_config to carry quirks and assign them when you add a port or start
it (wherever it's appropriated)?

-- 
With Best Regards,
Andy Shevchenko