Re: [PATCH v10 5/5] usb: dwc3: core: cleanup IRQ resources

From: Roger Quadros
Date: Fri Jun 10 2016 - 07:46:49 EST


On 10/06/16 14:44, Sergei Shtylyov wrote:
> On 6/10/2016 2:35 PM, Roger Quadros wrote:
>> On 10/06/16 13:39, Sergei Shtylyov wrote:
>>> Hello.
>>>
>>> On 6/10/2016 12:56 PM, Roger Quadros wrote:
>>>
>>>> Implementations might use different IRQs for
>>>> host, gadget so use named interrupt resources
>>>> to allow device tree to specify the interrupts.
>>>>
>>>> Following are the interrupt names
>>>>
>>>> Peripheral Interrupt - peripheral
>>>> HOST Interrupt - host
>>>>
>>>> Maintain backward compatibility for a single named
>>>> interrupt ("dwc3_usb3") for all interrupts as well as
>>>> unnamed interrupt at index 0 for all interrupts.
>>>>
>>>> As platform_get_irq_() variants are used, tackle
>>>
>>> platform_get_irq().
>>
>> OK.
>>>
>>>> the -EPROBE_DEFER case as well.
>>>>
>>>> Signed-off-by: Roger Quadros <rogerq@xxxxxx>
>>>> ---
>>>> v10:
>>>> - don't mention otg irq since we are not using it yet
>>>> - use platform_get_irq() and friends and check -EPROBE_DEFER case.
>>>>
>>>> drivers/usb/dwc3/core.c | 22 ++++++++--------------
>>>> drivers/usb/dwc3/gadget.c | 29 ++++++++++++++++++++++++++---
>>>> drivers/usb/dwc3/host.c | 41 ++++++++++++++++++++++++++++++++++++++++-
>>>> 3 files changed, 74 insertions(+), 18 deletions(-)
>>>>
>>>> diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
>>>> index 8fceeb1..131e7eb 100644
>>>> --- a/drivers/usb/dwc3/core.c
>>>> +++ b/drivers/usb/dwc3/core.c
>>> [...]
>>>> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
>>>> index 0f6fb8e..774a0d8 100644
>>>> --- a/drivers/usb/dwc3/gadget.c
>>>> +++ b/drivers/usb/dwc3/gadget.c
>>> [...]
>>>> @@ -2866,7 +2865,31 @@ static irqreturn_t dwc3_interrupt(int irq, void *_evt)
>>>> */
>>>> int dwc3_gadget_init(struct dwc3 *dwc)
>>>> {
>>>> - int ret;
>>>> + int ret, irq;
>>>> + struct platform_device *dwc3_pdev = to_platform_device(dwc->dev);
>>>> +
>>>> + irq = platform_get_irq_byname(dwc3_pdev, "peripheral");
>>>> + if (irq == -EPROBE_DEFER)
>>>> + return irq;
>>>> +
>>>> + if (irq <= 0) {
>>>> + irq = platform_get_irq_byname(dwc3_pdev, "dwc_usb3");
>>>> + if (irq == -EPROBE_DEFER)
>>>> + return irq;
>>>> +
>>>> + if (irq <= 0) {
>>>> + irq = platform_get_irq(dwc3_pdev, 0);
>>>> + if (irq <= 0) {
>>>> + if (irq != -EPROBE_DEFER) {
>>>> + dev_err(dwc->dev,
>>>> + "missing peripheral IRQ\n");
>>>> + }
>>>> + return irq;
>>>
>>> Iff irq == 0, you'll return success despite IRQ was "invalid". Was that intended?
>>
>> good catch. It wasn't intended. I guess i'll return -EINVAL then?
>>
>>>
>>>> + }
>>>> + }
>>>> + }
>>>> +
>>>> + dwc->irq_gadget = irq;
>>>>
>>>> dwc->ctrl_req = dma_alloc_coherent(dwc->dev, sizeof(*dwc->ctrl_req),
>>>> &dwc->ctrl_req_addr, GFP_KERNEL);
>>>> diff --git a/drivers/usb/dwc3/host.c b/drivers/usb/dwc3/host.c
>>>> index c679f63..eb5e8f9 100644
>>>> --- a/drivers/usb/dwc3/host.c
>>>> +++ b/drivers/usb/dwc3/host.c
>>>> @@ -24,7 +24,46 @@ int dwc3_host_init(struct dwc3 *dwc)
>>>> {
>>>> struct platform_device *xhci;
>>>> struct usb_xhci_pdata pdata;
>>>> - int ret;
>>>> + int ret, irq;
>>>> + struct resource *res;
>>>> + struct platform_device *dwc3_pdev = to_platform_device(dwc->dev);
>>>> +
>>>> + irq = platform_get_irq_byname(dwc3_pdev, "host");
>>>> + if (irq == -EPROBE_DEFER)
>>>> + return irq;
>>>> +
>>>> + if (irq <= 0) {
>>>> + irq = platform_get_irq_byname(dwc3_pdev, "dwc_usb3");
>>>> + if (irq == -EPROBE_DEFER)
>>>> + return irq;
>>>> +
>>>> + if (irq <= 0) {
>>>> + irq = platform_get_irq(dwc3_pdev, 0);
>>>> + if (irq <= 0) {
>>>> + if (irq != -EPROBE_DEFER) {
>>>> + dev_err(dwc->dev,
>>>> + "missing host IRQ\n");
>>>> + }
>>>> + return irq;
>>>
>>> Iff irq == 0, you'll return success despite IRQ was "invalid". Was that intended?
>
> I'd just consider 0 a valid IRQ, that's simpler. FYI, I've submitted to Greg KH a patch fixing platform_get_irq[_byname]() to not return 0 on failure. No reaction so far...

Maybe till your patch is in we can't really differentiate if it is error or not
so it is safer to consider it as error IMO.

cheers,
-roger