Re: [PATCH v14 3/6] usb: dwc3: Resize TX FIFOs to meet EP bursting requirements

[Wesley Cheng <wcheng@codeaurora.org>]

On 7/13/2021 11:40 PM, Felipe Balbi wrote:
> 
> Hi,
> 
> Thinh Nguyen <Thinh.Nguyen@synopsys.com> writes:
>>> Thinh Nguyen <Thinh.Nguyen@synopsys.com> writes:
>>>> Wesley Cheng wrote:
>>>>> Some devices have USB compositions which may require multiple endpoints
>>>>> that support EP bursting.  HW defined TX FIFO sizes may not always be
>>>>> sufficient for these compositions.  By utilizing flexible TX FIFO
>>>>> allocation, this allows for endpoints to request the required FIFO depth to
>>>>> achieve higher bandwidth.  With some higher bMaxBurst configurations, using
>>>>> a larger TX FIFO size results in better TX throughput.
>>>>>
>>>>> By introducing the check_config() callback, the resizing logic can fetch
>>>>> the maximum number of endpoints used in the USB composition (can contain
>>>>> multiple configurations), which helps ensure that the resizing logic can
>>>>> fulfill the configuration(s), or return an error to the gadget layer
>>>>> otherwise during bind time.
>>>>>
>>>>> Signed-off-by: Wesley Cheng <wcheng@codeaurora.org>
>>>>> ---
>>>>>  drivers/usb/dwc3/core.c   |  15 +++
>>>>>  drivers/usb/dwc3/core.h   |  16 ++++
>>>>>  drivers/usb/dwc3/ep0.c    |   2 +
>>>>>  drivers/usb/dwc3/gadget.c | 232 ++++++++++++++++++++++++++++++++++++++++++++++
>>>>>  4 files changed, 265 insertions(+)
>>>>>
>>>>> diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
>>>>> index ba74ad7..b194aecd 100644
>>>>> --- a/drivers/usb/dwc3/core.c
>>>>> +++ b/drivers/usb/dwc3/core.c
>>>>> @@ -1267,6 +1267,7 @@ static void dwc3_get_properties(struct dwc3 *dwc)
>>>>>  	u8			rx_max_burst_prd;
>>>>>  	u8			tx_thr_num_pkt_prd;
>>>>>  	u8			tx_max_burst_prd;
>>>>> +	u8			tx_fifo_resize_max_num;
>>>>>  	const char		*usb_psy_name;
>>>>>  	int			ret;
>>>>>  
>>>>> @@ -1282,6 +1283,13 @@ static void dwc3_get_properties(struct dwc3 *dwc)
>>>>>  	 */
>>>>>  	hird_threshold = 12;
>>>>>  
>>>>> +	/*
>>>>> +	 * default to a TXFIFO size large enough to fit 6 max packets.  This
>>>>> +	 * allows for systems with larger bus latencies to have some headroom
>>>>> +	 * for endpoints that have a large bMaxBurst value.
>>>>> +	 */
>>>>> +	tx_fifo_resize_max_num = 6;
>>>>> +
>>>>>  	dwc->maximum_speed = usb_get_maximum_speed(dev);
>>>>>  	dwc->max_ssp_rate = usb_get_maximum_ssp_rate(dev);
>>>>>  	dwc->dr_mode = usb_get_dr_mode(dev);
>>>>> @@ -1325,6 +1333,11 @@ static void dwc3_get_properties(struct dwc3 *dwc)
>>>>>  				&tx_thr_num_pkt_prd);
>>>>>  	device_property_read_u8(dev, "snps,tx-max-burst-prd",
>>>>>  				&tx_max_burst_prd);
>>>>> +	dwc->do_fifo_resize = device_property_read_bool(dev,
>>>>> +							"tx-fifo-resize");
>>>>> +	if (dwc->do_fifo_resize)
>>>>> +		device_property_read_u8(dev, "tx-fifo-max-num",
>>>>> +					&tx_fifo_resize_max_num);
>>>>
>>>> Why is this check here? The dwc->tx_fifo_resize_max_num should store
>>>> whatever property the user sets. Whether the driver wants to use this
>>>
>>> Ack!
>>>
>>>> property should depend on "dwc->do_fifo_resize". Also why don't we have
>>>> "snps," prefix to be consistent with the other properties?
>>>
>>> Ack!
>>>
>>>> Can we enforce to a single property? If the designer wants to enable
>>>> this feature, he/she can to provide the tx-fifo-max-num. This would
>>>> simplify the driver a bit. Since this is to optimize for performance,
>>>> the user should know/want/test the specific value if they want to set
>>>> for their setup and not hoping that the default setting not break their
>>>> setup. So we can remove the "do_fifo_resize" property and just check
>>>> whether tx_fifo_resize_max_num is set.
>>>
>>> Ack!
>>>
>>> All very valid points :-)
>>>

Hi Thinh/Felipe,

>>
>> Looks like this series already landed in Greg's testing branch. Not sure
>> how we usually handle this to address some of our concerns. Add fix
>> patches on top of Greg's testing branch?
> 
> yup, no choice anymore :-(
> 

Let me review your feedback, which had some good points.  We can add a
change addressing everything on top of what is merged on Greg's branch.
 Thanks for all the input!

Thanks
Wesley Cheng

-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project