linux-kernel - Re: [PATCH 2/2] bus: sunxi-rsb: Support atomic transfers

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: <4779430.31r3eYUQgx@jernej-laptop>
Date:   Sat, 05 Nov 2022 20:46:47 +0100
From:   Jernej Škrabec <jernej.skrabec@...il.com>
To:     Chen-Yu Tsai <wens@...e.org>, Samuel Holland <samuel@...lland.org>
Cc:     Ivaylo Dimitrov <ivo.g.dimitrov.75@...il.com>,
        linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        linux-sunxi@...ts.linux.dev, Samuel Holland <samuel@...lland.org>
Subject: Re: [PATCH 2/2] bus: sunxi-rsb: Support atomic transfers

Dne sobota, 05. november 2022 ob 20:19:53 CET je Samuel Holland napisal(a):
> When communicating with a PMIC during system poweroff (pm_power_off()),
> IRQs are disabled and we are in a RCU read-side critical section, so we
> cannot use wait_for_completion_io_timeout(). Instead, poll the status
> register for transfer completion.
> 
> Signed-off-by: Samuel Holland <samuel@...lland.org>
> ---
> 
>  drivers/bus/sunxi-rsb.c | 23 +++++++++++++++++------
>  1 file changed, 17 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/bus/sunxi-rsb.c b/drivers/bus/sunxi-rsb.c
> index 17343cd75338..0f0e498d4379 100644
> --- a/drivers/bus/sunxi-rsb.c
> +++ b/drivers/bus/sunxi-rsb.c
> @@ -267,6 +267,9 @@ EXPORT_SYMBOL_GPL(sunxi_rsb_driver_register);
>  /* common code that starts a transfer */
>  static int _sunxi_rsb_run_xfer(struct sunxi_rsb *rsb)
>  {
> +	bool timeout;
> +	u32 status;
> +
>  	if (readl(rsb->regs + RSB_CTRL) & RSB_CTRL_START_TRANS) {
>  		dev_dbg(rsb->dev, "RSB transfer still in progress\n");
>  		return -EBUSY;
> @@ -279,8 +282,16 @@ static int _sunxi_rsb_run_xfer(struct sunxi_rsb *rsb)
>  	writel(RSB_CTRL_START_TRANS | RSB_CTRL_GLOBAL_INT_ENB,
>  	       rsb->regs + RSB_CTRL);
> 
> -	if (!wait_for_completion_io_timeout(&rsb->complete,
> -					    
msecs_to_jiffies(100))) {
> +	if (irqs_disabled()) {
> +		timeout = readl_poll_timeout_atomic(rsb->regs + 
RSB_INTS,
> +						    status, 
status, 10, 100000);

It would be good to check only for RSB_INTS_LOAD_BSY, RSB_INTS_TRANS_ERR and 
RSB_INTS_TRANS_OVER flags and clear them afterwards. That way we avoid problems 
if this path is used outside power off case.

Best regards,
Jernej

> +	} else {
> +		timeout = !wait_for_completion_io_timeout(&rsb-
>complete,
> +							  
msecs_to_jiffies(100));
> +		status = rsb->status;
> +	}
> +
> +	if (timeout) {
>  		dev_dbg(rsb->dev, "RSB timeout\n");
> 
>  		/* abort the transfer */
> @@ -292,18 +303,18 @@ static int _sunxi_rsb_run_xfer(struct sunxi_rsb *rsb)
>  		return -ETIMEDOUT;
>  	}
> 
> -	if (rsb->status & RSB_INTS_LOAD_BSY) {
> +	if (status & RSB_INTS_LOAD_BSY) {
>  		dev_dbg(rsb->dev, "RSB busy\n");
>  		return -EBUSY;
>  	}
> 
> -	if (rsb->status & RSB_INTS_TRANS_ERR) {
> -		if (rsb->status & RSB_INTS_TRANS_ERR_ACK) {
> +	if (status & RSB_INTS_TRANS_ERR) {
> +		if (status & RSB_INTS_TRANS_ERR_ACK) {
>  			dev_dbg(rsb->dev, "RSB slave nack\n");
>  			return -EINVAL;
>  		}
> 
> -		if (rsb->status & RSB_INTS_TRANS_ERR_DATA) {
> +		if (status & RSB_INTS_TRANS_ERR_DATA) {
>  			dev_dbg(rsb->dev, "RSB transfer data 
error\n");
>  			return -EIO;
>  		}