linux-kernel - Re: [PATCH REVIEW 1/2] ASoC: codecs: lpass-rx-macro: fix RXn(rx,n) macro for DSM

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: <czlx4thp7thnb6jrauilpbtzgbq637rmnwlpifxq5b5jfa3lqm@toyy3b2viscr>
Date: Wed, 25 Sep 2024 11:27:54 +0300
From: Dmitry Baryshkov <dmitry.baryshkov@...aro.org>
To: Alexey Klimov <alexey.klimov@...aro.org>
Cc: srinivas.kandagatla@...aro.org, a39.skl@...il.com, 
	linux-sound@...r.kernel.org, lgirdwood@...il.com, broonie@...nel.org, perex@...ex.cz, 
	tiwai@...e.com, alsa-devel@...a-project.org, linux-arm-msm@...r.kernel.org, 
	linux-kernel@...r.kernel.org, krzysztof.kozlowski@...aro.org, vkoul@...nel.org, 
	klimov.linux@...il.com
Subject: Re: [PATCH REVIEW 1/2] ASoC: codecs: lpass-rx-macro: fix RXn(rx,n)
 macro for DSM_CTL and SEC7 regs

On Wed, Sep 25, 2024 at 05:38:22AM GMT, Alexey Klimov wrote:
> Turns out some registers of pre-2.5 version of rxmacro codecs are not
> located at the expected offsets but 0xc further away in memory.
> So far the detected registers are CDC_RX_RX2_RX_PATH_SEC7 and
> CDC_RX_RX2_RX_PATH_DSM_CTL.
> 
> CDC_RX_RXn_RX_PATH_DSM_CTL(rx, n) macro incorrectly generates the address
> 0x540 for RX2 but it should be 0x54C and it also overwrites
> CDC_RX_RX2_RX_PATH_SEC7 which is located at 0x540.
> The same goes for CDC_RX_RXn_RX_PATH_SEC7(rx, n).
> 
> Fix this by introducing additional rxn_reg_stride2 offset. For 2.5 version
> and above this offset will be equal to 0.
> With such change the corresponding RXn() macros will generate the same
> values for 2.5 codec version for all RX paths and the same old values
> for pre-2.5 version for RX0 and RX1. However for the latter case with RX2 path
> it will also add 2 * rxn_reg_stride2 on top.
> 
> Signed-off-by: Alexey Klimov <alexey.klimov@...aro.org>
> ---
>  sound/soc/codecs/lpass-rx-macro.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/sound/soc/codecs/lpass-rx-macro.c b/sound/soc/codecs/lpass-rx-macro.c
> index 71e0d3bffd3f..9288ddb705fe 100644
> --- a/sound/soc/codecs/lpass-rx-macro.c
> +++ b/sound/soc/codecs/lpass-rx-macro.c
> @@ -202,12 +202,14 @@
>  #define CDC_RX_RXn_RX_PATH_SEC3(rx, n)	(0x042c  + rx->rxn_reg_stride * n)
>  #define CDC_RX_RX0_RX_PATH_SEC4		(0x0430)
>  #define CDC_RX_RX0_RX_PATH_SEC7		(0x0434)
> -#define CDC_RX_RXn_RX_PATH_SEC7(rx, n)	(0x0434  + rx->rxn_reg_stride * n)
> +#define CDC_RX_RXn_RX_PATH_SEC7(rx, n)		\
> +	(0x0434 + rx->rxn_reg_stride * n + n * (n - 1) * rx->rxn_reg_stride2)

This is a nice hack to rule out n=0 and n=1, but maybe we can be more
obvious here:

(0x0434 + stride * n + (n > 2) ? stride2 : 0)

>  #define CDC_RX_DSM_OUT_DELAY_SEL_MASK	GENMASK(2, 0)
>  #define CDC_RX_DSM_OUT_DELAY_TWO_SAMPLE	0x2
>  #define CDC_RX_RX0_RX_PATH_MIX_SEC0	(0x0438)
>  #define CDC_RX_RX0_RX_PATH_MIX_SEC1	(0x043C)
> -#define CDC_RX_RXn_RX_PATH_DSM_CTL(rx, n)	(0x0440  + rx->rxn_reg_stride * n)
> +#define CDC_RX_RXn_RX_PATH_DSM_CTL(rx, n)	\
> +	(0x0440 + rx->rxn_reg_stride * n + n * (n - 1) * rx->rxn_reg_stride2)
>  #define CDC_RX_RXn_DSM_CLK_EN_MASK	BIT(0)
>  #define CDC_RX_RX0_RX_PATH_DSM_CTL	(0x0440)
>  #define CDC_RX_RX0_RX_PATH_DSM_DATA1	(0x0444)

-- 
With best wishes
Dmitry