linux-kernel - Re: [PATCH] memory/samsung: Maybe wrong triming parameter

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:   Thu, 7 May 2020 16:42:46 +0100
From:   Lukasz Luba <lukasz.luba@....com>
To:     Bernard Zhao <bernard@...o.com>, Kukjin Kim <kgene@...nel.org>,
        Krzysztof Kozlowski <krzk@...nel.org>,
        linux-pm@...r.kernel.org, linux-samsung-soc@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Cc:     opensource.kernel@...o.com
Subject: Re: [PATCH] memory/samsung: Maybe wrong triming parameter

Hi Bernard,


On 5/7/20 12:45 PM, Bernard Zhao wrote:
> In function create_timings_aligned, all the max is to use
> dmc->min_tck->xxx, aligned with val dmc->timings->xxx.
> But the dmc->timings->tFAW use dmc->min_tck->tXP?
> Maybe this point is wrong parameter useing.
> 
> Signed-off-by: Bernard Zhao <bernard@...o.com>
> ---
>   drivers/memory/samsung/exynos5422-dmc.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/memory/samsung/exynos5422-dmc.c b/drivers/memory/samsung/exynos5422-dmc.c
> index 81a1b1d01683..22a43d662833 100644
> --- a/drivers/memory/samsung/exynos5422-dmc.c
> +++ b/drivers/memory/samsung/exynos5422-dmc.c
> @@ -1091,7 +1091,7 @@ static int create_timings_aligned(struct exynos5_dmc *dmc, u32 *reg_timing_row,
>   	/* power related timings */
>   	val = dmc->timings->tFAW / clk_period_ps;
>   	val += dmc->timings->tFAW % clk_period_ps ? 1 : 0;
> -	val = max(val, dmc->min_tck->tXP);
> +	val = max(val, dmc->min_tck->tFAW);
>   	reg = &timing_power[0];
>   	*reg_timing_power |= TIMING_VAL2REG(reg, val);
>   
> 

Good catch! Indeed this should be a dmc->min_tck->tFAW used for
clamping.

It didn't show up in testing because the frequency values based on
which the 'clk_period_ps' are calculated are sane.
Check the dump below:

[    5.458227] DMC: mem tFAW=25000, clk_period_ps=6060
[    5.461743] DMC: tFAW=5, tXP=2 val=5
[    5.465273] DMC: mem tFAW=25000, clk_period_ps=4854
[    5.470101] DMC: tFAW=5, tXP=2 val=6
[    5.473668] DMC: mem tFAW=25000, clk_period_ps=3636
[    5.478507] DMC: tFAW=5, tXP=2 val=7
[    5.482072] DMC: mem tFAW=25000, clk_period_ps=2421
[    5.486951] DMC: tFAW=5, tXP=2 val=11
[    5.490531] DMC: mem tFAW=25000, clk_period_ps=1841
[    5.495439] DMC: tFAW=5, tXP=2 val=14
[    5.499113] DMC: mem tFAW=25000, clk_period_ps=1579
[    5.503877] DMC: tFAW=5, tXP=2 val=16
[    5.507476] DMC: mem tFAW=25000, clk_period_ps=1373
[    5.512368] DMC: tFAW=5, tXP=2 val=19
[    5.515968] DMC: mem tFAW=25000, clk_period_ps=1212
[    5.520826] DMC: tFAW=5, tXP=2 val=21

That's why in the existing configuration it does not harm
(the calculated 'val' is always >= 5) the board.

But I think this patch should be applied (after small changes in the
commit message).

@Krzysztof could you have a look on the commit message or take the
patch with small adjustment in the description, please?

I conditionally give (because of this description):

Reviewed-by: Lukasz Luba <lukasz.luba@....com>


Thank you Bernard for reporting and fixing this.

Regards,
Lukasz