usb: gadget: r8a66597: Fix a possible sleep-in-atomic-context bugs in r8a66597_queue()

The driver may sleep in an interrupt handler.
The function call path (from bottom to top) in Linux-4.16.7 is:

[FUNC] r8a66597_queue(GFP_KERNEL)
drivers/usb/gadget/udc/r8a66597-udc.c, 1193:
		r8a66597_queue in get_status
drivers/usb/gadget/udc/r8a66597-udc.c, 1301:
		get_status in setup_packet
drivers/usb/gadget/udc/r8a66597-udc.c, 1381:
		setup_packet in irq_control_stage
drivers/usb/gadget/udc/r8a66597-udc.c, 1508:
		irq_control_stage in r8a66597_irq (interrupt handler)

To fix this bug, GFP_KERNEL is replaced with GFP_ATOMIC.

This bug is found by my static analysis tool (DSAC-2) and checked by
my code review.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>

diff --git a/drivers/usb/gadget/udc/r8a66597-udc.c b/drivers/usb/gadget/udc/r8a66597-udc.c
index 24ee9867964b1ab6af9fece78d48dac8d6dfd82e..11e25a3f4f1fa86ea42677d7efbbdd1f0f83e6b5 100644 (file)
--- a/drivers/usb/gadget/udc/r8a66597-udc.c
+++ b/drivers/usb/gadget/udc/r8a66597-udc.c
@@ -1190,7 +1190,7 @@ __acquires(r8a66597->lock)
        r8a66597->ep0_req->length = 2;
        /* AV: what happens if we get called again before that gets through? */
        spin_unlock(&r8a66597->lock);
-       r8a66597_queue(r8a66597->gadget.ep0, r8a66597->ep0_req, GFP_KERNEL);
+       r8a66597_queue(r8a66597->gadget.ep0, r8a66597->ep0_req, GFP_ATOMIC);
        spin_lock(&r8a66597->lock);
 }