[U-Boot] [PATCH] usb: dwc3: gadget: make cache-maintenance on event buffers more robust

[Felipe Balbi]

Hi,

Marek Vasut <marex at denx.de> writes:
>> diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
>> index b2c7eb1..f58c7ba 100644
>> --- a/drivers/usb/dwc3/core.c
>> +++ b/drivers/usb/dwc3/core.c
>> @@ -125,6 +125,8 @@ static struct dwc3_event_buffer *dwc3_alloc_one_event_buffer(struct dwc3 *dwc,
>>  	if (!evt->buf)
>>  		return ERR_PTR(-ENOMEM);
>>  
>> +	dwc3_flush_cache((long)evt->buf, evt->length);
>> +
>
> Is the length aligned ? If not, you will get cache alignment warning.
> Also, address should be uintptr_t to avoid 32/64 bit issues .

if it's not aligned to 128 bits (at least, IIRC), DWC3 won't work.

>> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
>> index 1156662..61af71b 100644
>> --- a/drivers/usb/dwc3/gadget.c
>> +++ b/drivers/usb/dwc3/gadget.c
>> @@ -2668,11 +2668,12 @@ void dwc3_gadget_uboot_handle_interrupt(struct dwc3 *dwc)
>>  		int i;
>>  		struct dwc3_event_buffer *evt;
>>  
>> +		dwc3_thread_interrupt(0, dwc);
>> +
>> +		/* Clean + Invalidate the buffers after touching them */
>>  		for (i = 0; i < dwc->num_event_buffers; i++) {
>>  			evt = dwc->ev_buffs[i];
>>  			dwc3_flush_cache((long)evt->buf, evt->length);
>>  		}
>> -
>
> This makes me wonder, don't you need to invalidate the event buffer
> somewhere so that the new data would be fetched from RAM ?

nope. In linux we allocate from coherent

-- 
balbi