[PATCH] USB: usbip: clean up mixed use of _irq() and _irqsave()

[PATCH] USB: usbip: clean up mixed use of _irq() and _irqsave()

[Dan Carpenter]

It generally doesn't make sense to use _irq() and _irqsave() in the same
function because either some of the callers have disabled IRQs or they
haven't.  In this case, the v_recv_cmd_submit() appears to always be
called with IRQs enabled so the code works fine.  That means I could
convert it to either _irq() or _irqsave() but I chose to use _irqsave()
because it's more conservative and easier to review.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
 drivers/usb/usbip/vudc_rx.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/usb/usbip/vudc_rx.c b/drivers/usb/usbip/vudc_rx.c
index 1e8a23d92cb4..d4a2f30a7580 100644
--- a/drivers/usb/usbip/vudc_rx.c
+++ b/drivers/usb/usbip/vudc_rx.c
@@ -104,18 +104,18 @@ static int v_recv_cmd_submit(struct vudc *udc,
 	if (pdu->base.direction == USBIP_DIR_IN)
 		address |= USB_DIR_IN;
 
-	spin_lock_irq(&udc->lock);
+	spin_lock_irqsave(&udc->lock, flags);
 	urb_p->ep = vudc_find_endpoint(udc, address);
 	if (!urb_p->ep) {
 		/* we don't know the type, there may be isoc data! */
 		dev_err(&udc->pdev->dev, "request to nonexistent endpoint");
-		spin_unlock_irq(&udc->lock);
+		spin_unlock_irqrestore(&udc->lock, flags);
 		usbip_event_add(&udc->ud, VUDC_EVENT_ERROR_TCP);
 		ret = -EPIPE;
 		goto free_urbp;
 	}
 	urb_p->type = urb_p->ep->type;
-	spin_unlock_irq(&udc->lock);
+	spin_unlock_irqrestore(&udc->lock, flags);
 
 	urb_p->new = 1;
 	urb_p->seqnum = pdu->base.seqnum;
-- 
2.35.1

Re: [PATCH] USB: usbip: clean up mixed use of _irq() and _irqsave()

[Shuah Khan]

On 5/25/22 6:27 AM, Dan Carpenter wrote:
> It generally doesn't make sense to use _irq() and _irqsave() in the same
> function because either some of the callers have disabled IRQs or they
> haven't.  In this case, the v_recv_cmd_submit() appears to always be
> called with IRQs enabled so the code works fine.  That means I could
> convert it to either _irq() or _irqsave() but I chose to use _irqsave()
> because it's more conservative and easier to review.
> 
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> ---
>   drivers/usb/usbip/vudc_rx.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/usb/usbip/vudc_rx.c b/drivers/usb/usbip/vudc_rx.c
> index 1e8a23d92cb4..d4a2f30a7580 100644
> --- a/drivers/usb/usbip/vudc_rx.c
> +++ b/drivers/usb/usbip/vudc_rx.c
> @@ -104,18 +104,18 @@ static int v_recv_cmd_submit(struct vudc *udc,
>   	if (pdu->base.direction == USBIP_DIR_IN)
>   		address |= USB_DIR_IN;
>   
> -	spin_lock_irq(&udc->lock);
> +	spin_lock_irqsave(&udc->lock, flags);
>   	urb_p->ep = vudc_find_endpoint(udc, address);
>   	if (!urb_p->ep) {
>   		/* we don't know the type, there may be isoc data! */
>   		dev_err(&udc->pdev->dev, "request to nonexistent endpoint");
> -		spin_unlock_irq(&udc->lock);
> +		spin_unlock_irqrestore(&udc->lock, flags);
>   		usbip_event_add(&udc->ud, VUDC_EVENT_ERROR_TCP);
>   		ret = -EPIPE;
>   		goto free_urbp;
>   	}
>   	urb_p->type = urb_p->ep->type;
> -	spin_unlock_irq(&udc->lock);
> +	spin_unlock_irqrestore(&udc->lock, flags);
>   
>   	urb_p->new = 1;
>   	urb_p->seqnum = pdu->base.seqnum;
> 

Thank you for the patch.

Reviewed-by: Shuah Khan <skhan@linuxfoundation.org>

Greg, please pick this up.

thanks,
-- Shuah