Hi,

Richard Cochran [off-list ref] writes:

quoted

@@ -139,11 +141,24 @@ long ptp_ioctl(struct posix_clock *pc, unsigned int cmd, unsigned long arg)
 		break;
 
 	case PTP_EXTTS_REQUEST:
+	case PTP_EXTTS_REQUEST2:
+		memset(&req, 0, sizeof(req));
+
 		if (copy_from_user(&req.extts, (void __user *)arg,
 				   sizeof(req.extts))) {
 			err = -EFAULT;
 			break;
 		}
+		if ((req.extts.flags || req.extts.rsv[0] || req.extts.rsv[1])
+			&& cmd == PTP_EXTTS_REQUEST2) {
+			err = -EINVAL;
+			break;
+		} else if (cmd == PTP_EXTTS_REQUEST) {
+			req.extts.flags = 0;

This still isn't quite right.  Sorry that was my fault.

The req.extts.flags can be (PTP_ENABLE_FEATURE | PTP_RISING_EDGE |
PTP_FALLING_EDGE), and ENABLE is used immediately below in this case.

Please #define those bits into a valid mask, and then:

- for PTP_EXTTS_REQUEST2 check that ~mask is zero, and
- for PTP_EXTTS_REQUEST clear the ~mask bits for the drivers. 

Thanks again for cleaning this up!

good point. This will actually reduce the size of the patch 2.

-- 
balbi