linux-kernel - Re: [PATCH 2/7] staging: fsl-mc: use generic memory barriers

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:   Mon, 17 Jul 2017 14:38:01 +0100
From:   Robin Murphy <robin.murphy@....com>
To:     laurentiu.tudor@....com, gregkh@...uxfoundation.org,
        stuyoder@...il.com
Cc:     devel@...verdev.osuosl.org, ruxandra.radulescu@....com,
        arnd@...db.de, marc.zyngier@....com, roy.pledge@....com,
        linux-kernel@...r.kernel.org, agraf@...e.de,
        catalin.horghidan@....com, ioana.ciornei@....com,
        leoyang.li@....com, bharat.bhushan@....com,
        linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH 2/7] staging: fsl-mc: use generic memory barriers

On 17/07/17 14:26, laurentiu.tudor@....com wrote:
> From: Laurentiu Tudor <laurentiu.tudor@....com>
> 
> No need to use arch-specific memory barriers; switch to using generic
> ones. The rmb()s were useless so drop them.
> 
> Signed-off-by: Laurentiu Tudor <laurentiu.tudor@....com>
> ---
>  drivers/staging/fsl-mc/bus/mc-sys.c | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/staging/fsl-mc/bus/mc-sys.c b/drivers/staging/fsl-mc/bus/mc-sys.c
> index a1704c3..012abd5 100644
> --- a/drivers/staging/fsl-mc/bus/mc-sys.c
> +++ b/drivers/staging/fsl-mc/bus/mc-sys.c
> @@ -127,7 +127,8 @@ static inline void mc_write_command(struct mc_command __iomem *portal,
>  	/* copy command parameters into the portal */
>  	for (i = 0; i < MC_CMD_NUM_OF_PARAMS; i++)
>  		__raw_writeq(cmd->params[i], &portal->params[i]);
> -	__iowmb();
> +	/* ensure command params are committed before submitting it */
> +	wmb();
>  
>  	/* submit the command by writing the header */
>  	__raw_writeq(cmd->header, &portal->header);

AFAICS, just using writeq() here should ensure sufficient order against
the previous iomem accessors, without the need for explicit barriers.

Also, note that the __raw_*() variants aren't endian-safe, so consider
updating things to *_relaxed() where ordering doesn't matter.

Robin.

> @@ -150,9 +151,7 @@ static inline enum mc_cmd_status mc_read_response(struct mc_command __iomem *
>  	enum mc_cmd_status status;
>  
>  	/* Copy command response header from MC portal: */
> -	__iormb();
>  	resp->header = __raw_readq(&portal->header);
> -	__iormb();
>  	status = mc_cmd_hdr_read_status(resp);
>  	if (status != MC_CMD_STATUS_OK)
>  		return status;
> @@ -160,7 +159,6 @@ static inline enum mc_cmd_status mc_read_response(struct mc_command __iomem *
>  	/* Copy command response data from MC portal: */
>  	for (i = 0; i < MC_CMD_NUM_OF_PARAMS; i++)
>  		resp->params[i] = __raw_readq(&portal->params[i]);
> -	__iormb();
>  
>  	return status;
>  }
>