Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 30 Jun 2023 13:40:01 -0700
From:   Bjorn Andersson <quic_bjorande@quicinc.com>
To:     Komal Bajaj <quic_kbajaj@quicinc.com>
CC:     Andy Gross <agross@kernel.org>,
        Bjorn Andersson <andersson@kernel.org>,
        Konrad Dybcio <konrad.dybcio@linaro.org>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Srinivas Kandagatla <srinivas.kandagatla@linaro.org>,
        Conor Dooley <conor+dt@kernel.org>,
        <linux-arm-msm@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
        <devicetree@vger.kernel.org>
Subject: Re: [PATCH v4 5/6] soc: qcom: Add LLCC support for multi channel DDR
Message-ID: <20230630204001.GC1059662@hu-bjorande-lv.qualcomm.com>
References: <20230623141806.13388-1-quic_kbajaj@quicinc.com>
 <20230623141806.13388-6-quic_kbajaj@quicinc.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <20230623141806.13388-6-quic_kbajaj@quicinc.com>
Precedence: bulk

On Fri, Jun 23, 2023 at 07:48:05PM +0530, Komal Bajaj wrote:
> Add LLCC support for multi channel DDR configuration
> based on a feature register. Reading DDR channel
> confiuration uses nvmem framework, so select the
> dependency in Kconfig. Without this, there will be
> errors while building the driver with COMPILE_TEST only.

You may drop the last sentence, I don't think it's entirely correct.

> 
> Signed-off-by: Komal Bajaj <quic_kbajaj@quicinc.com>
> ---
>  drivers/soc/qcom/Kconfig     |  2 ++
>  drivers/soc/qcom/llcc-qcom.c | 33 ++++++++++++++++++++++++++++++---
>  2 files changed, 32 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/soc/qcom/Kconfig b/drivers/soc/qcom/Kconfig
> index a491718f8064..cc9ad41c63aa 100644
> --- a/drivers/soc/qcom/Kconfig
> +++ b/drivers/soc/qcom/Kconfig
> @@ -64,6 +64,8 @@ config QCOM_LLCC
>  	tristate "Qualcomm Technologies, Inc. LLCC driver"
>  	depends on ARCH_QCOM || COMPILE_TEST
>  	select REGMAP_MMIO
> +	select NVMEM
> +	select QCOM_SCM

I don't see anything your patch that warrants adding QCOM_SCM here,
is it needed, should it be a separate commit?

>  	help
>  	  Qualcomm Technologies, Inc. platform specific
>  	  Last Level Cache Controller(LLCC) driver for platforms such as,
> diff --git a/drivers/soc/qcom/llcc-qcom.c b/drivers/soc/qcom/llcc-qcom.c
> index 6cf373da5df9..3c29612da1c5 100644
> --- a/drivers/soc/qcom/llcc-qcom.c
> +++ b/drivers/soc/qcom/llcc-qcom.c
> @@ -12,6 +12,7 @@
>  #include <linux/kernel.h>
>  #include <linux/module.h>
>  #include <linux/mutex.h>
> +#include <linux/nvmem-consumer.h>
>  #include <linux/of.h>
>  #include <linux/of_device.h>
>  #include <linux/regmap.h>
> @@ -943,6 +944,19 @@ static int qcom_llcc_cfg_program(struct platform_device *pdev,
>  	return ret;
>  }
>  
> +static int qcom_llcc_get_cfg_index(struct platform_device *pdev, u8 *cfg_index)
> +{
> +	int ret;
> +
> +	ret = nvmem_cell_read_u8(&pdev->dev, "multi-chan-ddr", cfg_index);
> +	if (ret == -ENOENT) {
> +		*cfg_index = 0;
> +		return 0;
> +	}
> +
> +	return ret;
> +}
> +
>  static int qcom_llcc_remove(struct platform_device *pdev)
>  {
>  	/* Set the global pointer to a error code to avoid referencing it */
> @@ -975,11 +989,13 @@ static int qcom_llcc_probe(struct platform_device *pdev)
>  	struct device *dev = &pdev->dev;
>  	int ret, i;
>  	struct platform_device *llcc_edac;
> -	const struct qcom_llcc_config *cfg;
> +	const struct qcom_llcc_config *cfg, *entry;
>  	const struct llcc_slice_config *llcc_cfg;
>  	u32 sz;
> +	u8 cfg_index;
>  	u32 version;
>  	struct regmap *regmap;
> +	u32 num_entries = 0;
>  
>  	drv_data = devm_kzalloc(dev, sizeof(*drv_data), GFP_KERNEL);
>  	if (!drv_data) {
> @@ -1040,8 +1056,19 @@ static int qcom_llcc_probe(struct platform_device *pdev)
>  
>  	drv_data->version = version;
>  
> -	llcc_cfg = cfg[0]->sct_data;
> -	sz = cfg[0]->size;
> +	ret = qcom_llcc_get_cfg_index(pdev, &cfg_index);
> +	if (ret)
> +		goto err;
> +
> +	for (entry = cfg; entry->sct_data; entry++, num_entries++)
> +		;

This is still unnecessarily "clever":
"For each valid entry, do nothing, while incrementing num_entries".

How about just writing:
"For each valid entry, increment num_entries"

	for (entry = cfg; entry->sct_data; entry++)
		num_entries++;


> +	if (cfg_index >= num_entries || cfg_index < 0) {

cfg_index is an unsiged number, so it's unlikely to be negative.

Also, "cfg_index" and "num_entries" are values in the same domain, so
keeping their names related is beneficial - i.e. rename num_entries to
num_cfgs.

Regards,
Bjorn

> +		ret = -EINVAL;
> +		goto err;
> +	}
> +
> +	llcc_cfg = cfg[cfg_index].sct_data;
> +	sz = cfg[cfg_index].size;
>  
>  	for (i = 0; i < sz; i++)
>  		if (llcc_cfg[i].slice_id > drv_data->max_slices)
> -- 
> 2.40.1
>