Re: [PATCH V3 2/3] usb: dwc3: qcom: Add interconnect support in dwc3 driver

[Matthias Kaehlcke <mka@chromium.org>]

On Tue, Sep 17, 2019 at 04:40:48PM +0530, Chandana Kishori Chiluveru wrote:
> Add interconnect support in dwc3-qcom driver to vote for bus
> bandwidth.
> 
> This requires for two different paths - from USB master to
> DDR slave. The other is from APPS master to USB slave.
> 
> Signed-off-by: Chandana Kishori Chiluveru <cchiluve@codeaurora.org>
> ---
>  drivers/usb/dwc3/dwc3-qcom.c | 145 ++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 143 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/usb/dwc3/dwc3-qcom.c b/drivers/usb/dwc3/dwc3-qcom.c
> index 184df4d..2a2f5af 100644
> --- a/drivers/usb/dwc3/dwc3-qcom.c
> +++ b/drivers/usb/dwc3/dwc3-qcom.c
>
> ...
>
> +/**
> + * dwc3_qcom_interconnect_init() - Get interconnect path handles

nit: "Initialize the interconnect" or similar?

> + * @qcom:			Pointer to the concerned usb core.
> + *
> + */
> +static int dwc3_qcom_interconnect_init(struct dwc3_qcom *qcom)
> +{
> +	struct device *dev = qcom->dev;
> +	int ret;
> +
> +	qcom->usb_ddr_icc_path = of_icc_get(dev, "usb-ddr");
> +	if (IS_ERR(qcom->usb_ddr_icc_path)) {
> +		dev_err(dev, "Error: (%ld) failed getting usb-ddr path\n",
> +			PTR_ERR(qcom->usb_ddr_icc_path));
> +		return PTR_ERR(qcom->usb_ddr_icc_path);
> +	}
> +
> +	qcom->apps_usb_icc_path = of_icc_get(dev, "apps-usb");
> +	if (IS_ERR(qcom->apps_usb_icc_path)) {
> +		dev_err(dev, "Error: (%ld) failed getting apps-usb path\n",
> +				PTR_ERR(qcom->apps_usb_icc_path));
> +		return PTR_ERR(qcom->apps_usb_icc_path);
> +	}
> +
> +	ret = dwc3_qcom_interconnect_enable(qcom);
> +	if (ret) {
> +		dev_err(dev, "failed to enable interconnect %d\n", ret);

nit: 'interconnect: %d', otherwise the message could be read as
'failed to enable interconnect number N'.

> +		return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +/**
> + * dwc3_qcom_interconnect_exit() - Release interconnect path handles
> + * @qcom:			Pointer to the concerned usb core.
> + *
> + * This function is used to release interconnect path handle.
> + */
> +static void dwc3_qcom_interconnect_exit(struct dwc3_qcom *qcom)
> +{
> +	icc_put(qcom->usb_ddr_icc_path);
> +	icc_put(qcom->apps_usb_icc_path);
> +}
> +
> +/* Currently we only use bandwidth level, so just "enable" interconnects */
> +static int dwc3_qcom_interconnect_enable(struct dwc3_qcom *qcom)
> +{
> +	struct dwc3 *dwc;
> +	int ret;
> +
> +	dwc = platform_get_drvdata(qcom->dwc3);
> +	if (!dwc) {
> +		dev_err(qcom->dev, "Failed to get dwc3 device\n");
> +		return -EPROBE_DEFER;
> +	}

I understand the need for the/some check (see
https://patchwork.kernel.org/patch/11146903/#22885491 and my reply),
but I'm not convinced it should be done here. The function can be
called from other contexts than _probe(), so returning -EPROBE_DEFER
seems wrong, although I understand that you want _probe() to return
this value.

I would suggest to do this (or another) check early in
_probe(). Returning -EPROBE_DEFER from that context makes sense, and
it should be the only time the check is actually needed.

As commented on v2 I don't particularly like the idea of using a half
initialized struct (dwc3), even more when what is initialized or not
varies at runtime (since dwc3_probe() and qwc3_qcom_probe() appear to
run in parallel). IIUC device_is_bound() returns true when a device is
fully initialized (i.e. it's _probe() completed), I'd suggest to
evaluate to use it instead of checking the drvdata. In this case the
drvdata check should be ok, but in general these kind of races don't
seem a good idea, maybe tomorrow someone wants to access another
struct member, which isn't always initialized.

Others: please correct me if using half-initialized structs is
routine in kernel drivers or device_is_bound() is the wrong tool.

> +
> +	if (dwc->maximum_speed == USB_SPEED_SUPER) {
> +		ret = icc_set_bw(qcom->usb_ddr_icc_path,
> +			USB_MEMORY_AVG_SS_BW, USB_MEMORY_PEAK_SS_BW);
> +		if (ret)
> +			return ret;
> +	} else {
> +		ret = icc_set_bw(qcom->usb_ddr_icc_path,
> +			USB_MEMORY_AVG_HS_BW, USB_MEMORY_PEAK_HS_BW);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	ret = icc_set_bw(qcom->apps_usb_icc_path,
> +		APPS_USB_AVG_BW, APPS_USB_PEAK_BW);
> +	if (ret)
> +		goto err_disable_mem_path;
> +
> +	return 0;
> +
> +err_disable_mem_path:
> +	icc_set_bw(qcom->usb_ddr_icc_path, 0, 0);
> +
> +	return ret;
> +}
> +
> +/* To disable an interconnect, we just set its bandwidth to 0 */
> +static int dwc3_qcom_interconnect_disable(struct dwc3_qcom *qcom)
> +{
> +	struct dwc3 *dwc = platform_get_drvdata(qcom->dwc3);
> +	int ret;
> +
> +	ret = icc_set_bw(qcom->usb_ddr_icc_path, 0, 0);
> +	if (ret)
> +		return ret;
> +
> +	ret = icc_set_bw(qcom->apps_usb_icc_path, 0, 0);
> +	if (ret)
> +		goto err_reenable_memory_path;
> +
> +	return 0;
> +
> +	/* Re-enable things in the event of an error */
> +err_reenable_memory_path:
> +	if (dwc->maximum_speed == USB_SPEED_SUPER)
> +		icc_set_bw(qcom->usb_ddr_icc_path,
> +			USB_MEMORY_AVG_SS_BW, USB_MEMORY_PEAK_SS_BW);
> +	else
> +		icc_set_bw(qcom->usb_ddr_icc_path,
> +			USB_MEMORY_AVG_HS_BW, USB_MEMORY_PEAK_HS_BW);

This is essentially the same as in _interconnect_enable(). You could
consider a helper function (inline?) for 'enabling' the memory path,
which would make things more compact and also allow to get rid of the
goto:

if (ret)
	qcom_dwc3_interconnect_enable_mempath(qcom);

return ret;

> +
> +	return ret;
> +}
> +
>  static irqreturn_t qcom_dwc3_resume_irq(int irq, void *data)
>  {
>  	struct dwc3_qcom *qcom = data;
> @@ -494,6 +626,12 @@ static int dwc3_qcom_probe(struct platform_device *pdev)
>  		goto depopulate;
>  	}
>  
> +	ret = dwc3_qcom_interconnect_init(qcom);
> +	if (ret) {
> +		dev_err(dev, "failed to init interconnect handles %d\n", ret);

nit: remove 'handles' (and add ':' before the error code), it's an
implementation detail, and now _icc_init() also calls _icc_enable(),
hence it's not only the handles.