linux-kernel - Re: [PATCH] serial: sh-sci: Save and restore SCDL and SCCKS

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]

Message-ID: <289b6cba-b305-4de9-86a7-24520af16298@tuxon.dev>
Date: Wed, 19 Feb 2025 17:04:45 +0200
From: Claudiu Beznea <claudiu.beznea@...on.dev>
To: Geert Uytterhoeven <geert+renesas@...der.be>,
 Claudiu Beznea <claudiu.beznea.uj@...renesas.com>,
 Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
 Jiri Slaby <jirislaby@...nel.org>,
 Wolfram Sang <wsa+renesas@...g-engineering.com>
Cc: linux-serial@...r.kernel.org, linux-renesas-soc@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH] serial: sh-sci: Save and restore SCDL and SCCKS

Hi, Geert,

On 19.02.2025 16:24, Geert Uytterhoeven wrote:
> On (H)SCIF with a Baud Rate Generator for External Clock (BRG), there
> are multiple ways to configure the requested serial speed.  If firmware
> uses a different method than Linux, and if any debug info is printed
> after the Bit Rate Register (SCBRR) is restored, but before termios is
> reconfigured (which configures the alternative method), the system may
> lock-up during resume.
> 
> Fix this by saving and restoring the contents of the Frequency Division
> (DL) and Clock Select (CKS) registers as well.

Keeping the thinks RZ/G3S focused (as proposed in the first versions of
this support), I missed there might be other registers on other IP variants.

Reviewing the full list of registers from [1], maybe the HSSRR and
SCPCR should be saved/restored as well?

Thank you,
Claudiu


[1]
https://elixir.bootlin.com/linux/v6.13.3/source/drivers/tty/serial/sh-sci.h#L14

> 
> Fixes: 22a6984c5b5df8ea ("serial: sh-sci: Update the suspend/resume support")
> Signed-off-by: Geert Uytterhoeven <geert+renesas@...der.be>

For this patch:
Tested-by: Claudiu Beznea <claudiu.beznea.uj@...renesas.com>
Reviewed-by: Claudiu Beznea <claudiu.beznea.uj@...renesas.com>

> ---
> This can be reproduced on e.g. Salvator-X(S) by enabling the debug
> print in sci_brg_calc(), and using s2ram with no_console_suspend.
> ---
>  drivers/tty/serial/sh-sci.c | 10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
> index df6512c9c0ff28db..70f34b8a93888eb9 100644
> --- a/drivers/tty/serial/sh-sci.c
> +++ b/drivers/tty/serial/sh-sci.c
> @@ -109,6 +109,8 @@ struct sci_suspend_regs {
>  	u16 scscr;
>  	u16 scfcr;
>  	u16 scsptr;
> +	u16 scdl;
> +	u16 sccks;
>  	u8 scbrr;
>  	u8 semr;
>  };
> @@ -3571,6 +3573,10 @@ static void sci_console_save(struct sci_port *s)
>  		regs->scfcr = sci_serial_in(port, SCFCR);
>  	if (sci_getreg(port, SCSPTR)->size)
>  		regs->scsptr = sci_serial_in(port, SCSPTR);
> +	if (sci_getreg(port, SCDL)->size)
> +		regs->scdl = sci_serial_in(port, SCDL);
> +	if (sci_getreg(port, SCCKS)->size)
> +		regs->sccks = sci_serial_in(port, SCCKS);
>  	if (sci_getreg(port, SCBRR)->size)
>  		regs->scbrr = sci_serial_in(port, SCBRR);
>  	if (sci_getreg(port, SEMR)->size)
> @@ -3590,6 +3596,10 @@ static void sci_console_restore(struct sci_port *s)
>  		sci_serial_out(port, SCFCR, regs->scfcr);
>  	if (sci_getreg(port, SCSPTR)->size)
>  		sci_serial_out(port, SCSPTR, regs->scsptr);
> +	if (sci_getreg(port, SCDL)->size)
> +		sci_serial_out(port, SCDL, regs->scdl);
> +	if (sci_getreg(port, SCCKS)->size)
> +		sci_serial_out(port, SCCKS, regs->sccks);
>  	if (sci_getreg(port, SCBRR)->size)
>  		sci_serial_out(port, SCBRR, regs->scbrr);
>  	if (sci_getreg(port, SEMR)->size)