Re: [PATCH 3/3] drm/msm/hdmi: move resource allocation to probe function

[Dmitry Baryshkov <dmitry.baryshkov@linaro.org>]

On 24/08/2022 02:58, Abhinav Kumar wrote:
> 
> 
> On 6/16/2022 12:59 AM, Dmitry Baryshkov wrote:
>> Rather than having all resource allocation happen in the _bind function
>> (resulting in possible EPROBE_DEFER returns and component bind/unbind
>> cycles) allocate and check all resources in _probe function. While we
>> are at it, use platform_get_irq() to get the IRQ rather than going
>> through the irq_of_parse_and_map().
>>
>> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
>> ---
>>   drivers/gpu/drm/msm/hdmi/hdmi.c | 295 +++++++++++++++-----------------
>>   1 file changed, 134 insertions(+), 161 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/msm/hdmi/hdmi.c 
>> b/drivers/gpu/drm/msm/hdmi/hdmi.c
>> index 8dfe5690366b..429abd622c81 100644
>> --- a/drivers/gpu/drm/msm/hdmi/hdmi.c
>> +++ b/drivers/gpu/drm/msm/hdmi/hdmi.c
>> @@ -75,8 +75,6 @@ static void msm_hdmi_destroy(struct hdmi *hdmi)
>>       if (hdmi->i2c)
>>           msm_hdmi_i2c_destroy(hdmi->i2c);
>> -
>> -    platform_set_drvdata(hdmi->pdev, NULL);
> Do we still not need to do this in .remove?
>>   }
>>   static int msm_hdmi_get_phy(struct hdmi *hdmi)
>> @@ -116,138 +114,10 @@ static int msm_hdmi_get_phy(struct hdmi *hdmi)
>>    * we are to EPROBE_DEFER we want to do it here, rather than later
>>    * at modeset_init() time
>>    */
>> -static struct hdmi *msm_hdmi_init(struct platform_device *pdev)
>> +static int msm_hdmi_init(struct hdmi *hdmi)
>>   {
>> -    struct hdmi_platform_config *config = pdev->dev.platform_data;
>> -    struct hdmi *hdmi = NULL;
>> -    struct resource *res;
>> -    int i, ret;
>> -
>> -    hdmi = devm_kzalloc(&pdev->dev, sizeof(*hdmi), GFP_KERNEL);
>> -    if (!hdmi) {
>> -        ret = -ENOMEM;
>> -        goto fail;
>> -    }
>> -
>> -    hdmi->pdev = pdev;
>> -    hdmi->config = config;
>> -    spin_lock_init(&hdmi->reg_lock);
>> -
>> -    hdmi->mmio = msm_ioremap(pdev, "core_physical");
>> -    if (IS_ERR(hdmi->mmio)) {
>> -        ret = PTR_ERR(hdmi->mmio);
>> -        goto fail;
>> -    }
>> -
>> -    /* HDCP needs physical address of hdmi register */
>> -    res = platform_get_resource_byname(pdev, IORESOURCE_MEM,
>> -        "core_physical");
>> -    if (!res) {
>> -        ret = -EINVAL;
>> -        goto fail;
>> -    }
>> -    hdmi->mmio_phy_addr = res->start;
>> -
>> -    hdmi->qfprom_mmio = msm_ioremap(pdev, "qfprom_physical");
>> -    if (IS_ERR(hdmi->qfprom_mmio)) {
>> -        DRM_DEV_INFO(&pdev->dev, "can't find qfprom resource\n");
>> -        hdmi->qfprom_mmio = NULL;
>> -    }
>> -
>> -    hdmi->hpd_regs = devm_kcalloc(&pdev->dev,
>> -                      config->hpd_reg_cnt,
>> -                      sizeof(hdmi->hpd_regs[0]),
>> -                      GFP_KERNEL);
>> -    if (!hdmi->hpd_regs) {
>> -        ret = -ENOMEM;
>> -        goto fail;
>> -    }
>> -    for (i = 0; i < config->hpd_reg_cnt; i++)
>> -        hdmi->hpd_regs[i].supply = config->hpd_reg_names[i];
>> -
>> -    ret = devm_regulator_bulk_get(&pdev->dev, config->hpd_reg_cnt, 
>> hdmi->hpd_regs);
>> -    if (ret) {
>> -        DRM_DEV_ERROR(&pdev->dev, "failed to get hpd regulator: 
>> %d\n", ret);
>> -        goto fail;
>> -    }
>> -
>> -    hdmi->pwr_regs = devm_kcalloc(&pdev->dev,
>> -                      config->pwr_reg_cnt,
>> -                      sizeof(hdmi->pwr_regs[0]),
>> -                      GFP_KERNEL);
>> -    if (!hdmi->pwr_regs) {
>> -        ret = -ENOMEM;
>> -        goto fail;
>> -    }
>> -
>> -    for (i = 0; i < config->pwr_reg_cnt; i++)
>> -        hdmi->pwr_regs[i].supply = config->pwr_reg_names[i];
>> -
>> -    ret = devm_regulator_bulk_get(&pdev->dev, config->pwr_reg_cnt, 
>> hdmi->pwr_regs);
>> -    if (ret) {
>> -        DRM_DEV_ERROR(&pdev->dev, "failed to get pwr regulator: 
>> %d\n", ret);
>> -        goto fail;
>> -    }
>> -
>> -    hdmi->hpd_clks = devm_kcalloc(&pdev->dev,
>> -                      config->hpd_clk_cnt,
>> -                      sizeof(hdmi->hpd_clks[0]),
>> -                      GFP_KERNEL);
>> -    if (!hdmi->hpd_clks) {
>> -        ret = -ENOMEM;
>> -        goto fail;
>> -    }
>> -    for (i = 0; i < config->hpd_clk_cnt; i++) {
>> -        struct clk *clk;
>> -
>> -        clk = msm_clk_get(pdev, config->hpd_clk_names[i]);
>> -        if (IS_ERR(clk)) {
>> -            ret = PTR_ERR(clk);
>> -            DRM_DEV_ERROR(&pdev->dev, "failed to get hpd clk: %s 
>> (%d)\n",
>> -                    config->hpd_clk_names[i], ret);
>> -            goto fail;
>> -        }
>> -
>> -        hdmi->hpd_clks[i] = clk;
>> -    }
>> -
>> -    hdmi->pwr_clks = devm_kcalloc(&pdev->dev,
>> -                      config->pwr_clk_cnt,
>> -                      sizeof(hdmi->pwr_clks[0]),
>> -                      GFP_KERNEL);
>> -    if (!hdmi->pwr_clks) {
>> -        ret = -ENOMEM;
>> -        goto fail;
>> -    }
>> -    for (i = 0; i < config->pwr_clk_cnt; i++) {
>> -        struct clk *clk;
>> -
>> -        clk = msm_clk_get(pdev, config->pwr_clk_names[i]);
>> -        if (IS_ERR(clk)) {
>> -            ret = PTR_ERR(clk);
>> -            DRM_DEV_ERROR(&pdev->dev, "failed to get pwr clk: %s 
>> (%d)\n",
>> -                    config->pwr_clk_names[i], ret);
>> -            goto fail;
>> -        }
>> -
>> -        hdmi->pwr_clks[i] = clk;
>> -    }
>> -
>> -    hdmi->hpd_gpiod = devm_gpiod_get_optional(&pdev->dev, "hpd", 
>> GPIOD_IN);
>> -    /* This will catch e.g. -EPROBE_DEFER */
>> -    if (IS_ERR(hdmi->hpd_gpiod)) {
>> -        ret = PTR_ERR(hdmi->hpd_gpiod);
>> -        DRM_DEV_ERROR(&pdev->dev, "failed to get hpd gpio: (%d)\n", 
>> ret);
>> -        goto fail;
>> -    }
>> -
>> -    if (!hdmi->hpd_gpiod)
>> -        DBG("failed to get HPD gpio");
>> -
>> -    if (hdmi->hpd_gpiod)
>> -        gpiod_set_consumer_name(hdmi->hpd_gpiod, "HDMI_HPD");
>> -
>> -    devm_pm_runtime_enable(&pdev->dev);
>> +    struct platform_device *pdev = hdmi->pdev;
>> +    int ret;
> 
> What about the rest of the msm_hdmi_init() function?
> 
> msm_hdmi_i2c_init, msm_hdmi_get_phy and msm_hdmi_hdcp_init have been 
> left behind. Any reason for that?

msm_hdmi_i2c_init() allocates new adapter, so it should be part of bind().

msm_hdmi_hdcp_init() just allocates a chunk of memory (other actions are 
infallible). Also I did not want to move a piece of code that I can not 
really test.

As for the msm_hdmi_get_phy(), I don't remember why I didn't move it. 
But you are right, it makes sense to move it. I'll check it for v2.

> 
> 
>>       hdmi->workq = alloc_ordered_workqueue("msm_hdmi", 0);

-- 
With best wishes
Dmitry