Thu Mar 3 05:49:58 2022 UTC ()
usbnet: Don't waste time calling uno_stop if device is detaching.

The hardware is most likely gone, so trying to write to its registers
(and, in some cases, wait until a timeout for a device to reset) is a
waste of time.  Even if it was detached only in software with drvctl,
reattaching it will reset the device anyway.


(riastradh)
diff -r1.67 -r1.68 src/sys/dev/usb/usbnet.c
cvs diff -r1.67 -r1.68 src/sys/dev/usb/usbnet.c (expand / switch to context diff)
--- src/sys/dev/usb/usbnet.c 2022/03/03 05:49:44 1.67
+++ src/sys/dev/usb/usbnet.c 2022/03/03 05:49:58 1.68
@@ -1,4 +1,4 @@
-/*	$NetBSD: usbnet.c,v 1.67 2022/03/03 05:49:44 riastradh Exp $	*/
+/*	$NetBSD: usbnet.c,v 1.68 2022/03/03 05:49:58 riastradh Exp $	*/
 
 /*
  * Copyright (c) 2019 Matthew R. Green
@@ -31,7 +31,7 @@
  */
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: usbnet.c,v 1.67 2022/03/03 05:49:44 riastradh Exp $");
+__KERNEL_RCSID(0, "$NetBSD: usbnet.c,v 1.68 2022/03/03 05:49:58 riastradh Exp $");
 
 #include <sys/param.h>
 #include <sys/kernel.h>
@@ -1154,8 +1154,13 @@
 	 * the hardware.  The driver's uno_stop routine now has
 	 * exclusive access to any registers that might previously have
 	 * been used by to ifmedia, mii, or ioctl callbacks.
+	 *
+	 * Don't bother if the device is being detached, though -- if
+	 * it's been unplugged then there's no point in trying to touch
+	 * the registers.
 	 */
-	uno_stop(un, ifp, disable);
+	if (!unp->unp_dying)
+		uno_stop(un, ifp, disable);
 
 	/* Stop transfers. */
 	usbnet_ep_stop_pipes(un);