linux-kernel - Re: [PATCH 3/3] thermal/drivers/tsens: Extract and shift-in optional MSB

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: <ecf48eaf-b4b9-3749-894d-0f06d0b954e3@linaro.org>
Date:   Thu, 6 Apr 2023 20:47:11 +0300
From:   Dmitry Baryshkov <dmitry.baryshkov@...aro.org>
To:     Bryan O'Donoghue <bryan.odonoghue@...aro.org>, amitk@...nel.org,
        thara.gopinath@...il.com, agross@...nel.org, andersson@...nel.org,
        konrad.dybcio@...aro.org, rafael@...nel.org,
        daniel.lezcano@...aro.org, rui.zhang@...el.com
Cc:     linux-pm@...r.kernel.org, linux-arm-msm@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/3] thermal/drivers/tsens: Extract and shift-in optional
 MSB

On 06/04/2023 17:58, Bryan O'Donoghue wrote:
> In msm8939 some of the sensor calibration data traverses byte boundaries.
> Two examples of this are thermal sensor 2 point 1 and sensor 9 point 2.
> 
> For sensor 2 point 1 we can get away with a simple read traversing byte
> boundaries as the calibration most significant bits are adjacent to the
> least significant across the byte boundary.
> 
> In this case a read starting at the end of the first byte for nine bits
> will deliver up the data we want.
> 
> In the case of sensor 9 point 2 however, the most significant bits are not
> adjacent and so therefore we need to perform two reads and or the bits
> together.
> 
> If reg.p1_shift or reg.p2_shift is set then automatically search for
> pX_sY_msb in the dts applying pX_shift as a right shift or into the pX_sY
> value.

I think that having this in the common code is a bit of an overkill. No 
other platform has this 'peculiarity' up to now. So, it might be better 
to add 8939-specific calibration function that calls 
tsens_read_calibration(), mixes in the s10_p2_msb and then calls 
compute_intercept_slope().

> 
> Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@...aro.org>
> ---
>   drivers/thermal/qcom/tsens.c | 33 +++++++++++++++++++++++++++++++++
>   1 file changed, 33 insertions(+)
> 
> diff --git a/drivers/thermal/qcom/tsens.c b/drivers/thermal/qcom/tsens.c
> index a260f563b4889..eff2c8671c343 100644
> --- a/drivers/thermal/qcom/tsens.c
> +++ b/drivers/thermal/qcom/tsens.c
> @@ -74,6 +74,7 @@ int tsens_read_calibration(struct tsens_priv *priv, int shift, u32 *p1, u32 *p2,
>   {
>   	u32 mode;
>   	u32 base1, base2;
> +	u32 msb;
>   	char name[] = "sXX_pY_backup"; /* s10_p1_backup */
>   	int i, ret;
>   
> @@ -122,6 +123,22 @@ int tsens_read_calibration(struct tsens_priv *priv, int shift, u32 *p1, u32 *p2,
>   
>   		dev_dbg(priv->dev, "%s 0x%x\n", name, p1[i]);
>   
> +		if (priv->reg && priv->reg[i].p1_shift) {
> +			ret = snprintf(name, sizeof(name), "s%d_p1_msb",
> +				       priv->sensor[i].hw_id);
> +			if (ret < 0)
> +				return ret;
> +
> +			ret = nvmem_cell_read_variable_le_u32(priv->dev, name, &msb);
> +			if (ret) {
> +				dev_err(priv->dev, "Failed to read %s\n", name);
> +				return ret;
> +			}
> +
> +			dev_dbg(priv->dev, "%s 0x%x\n", name, msb);
> +			p1[i] |= msb >> priv->reg[i].p1_shift;
> +		}
> +
>   		ret = snprintf(name, sizeof(name), "s%d_p2%s", priv->sensor[i].hw_id,
>   			       backup ? "_backup" : "");
>   		if (ret < 0)
> @@ -134,6 +151,22 @@ int tsens_read_calibration(struct tsens_priv *priv, int shift, u32 *p1, u32 *p2,
>   		}
>   
>   		dev_dbg(priv->dev, "%s 0x%x\n", name, p2[i]);
> +
> +		if (priv->reg && priv->reg[i].p2_shift) {
> +			ret = snprintf(name, sizeof(name), "s%d_p2_msb",
> +				       priv->sensor[i].hw_id);
> +			if (ret < 0)
> +				return ret;
> +
> +			ret = nvmem_cell_read_variable_le_u32(priv->dev, name, &msb);
> +			if (ret) {
> +				dev_err(priv->dev, "Failed to read %s\n", name);
> +				return ret;
> +			}
> +
> +			dev_dbg(priv->dev, "%s 0x%x\n", name, msb);
> +			p2[i] |= msb >> priv->reg[i].p2_shift;
> +		}
>   	}
>   
>   	switch (mode) {

-- 
With best wishes
Dmitry