Re: [ANNOUNCE] udev 046 release

[Kay Sievers]

On Fri, 2004-11-19 at 12:26 +0100, Mathieu Segaud wrote:
> seems like these changes broke something in rules applying to eth* devices.
> the rules put and still working with udev 045 have no effect, now....
> not so inconvenient now that I've got just one card in my box, but I guess
> it could be a show-stopper for laptop users.
> 
> My rules which can be found at the end of /etc/udev/rules.d/50-udev.rules are:
> 
> KERNEL="eth*", SYSFS{address}="00:10:5a:49:36:d8", NAME="external"
> KERNEL="eth*", SYSFS{address}="00:50:04:69:db:56", NAME="private"
> KERNEL="eth*", SYSFS{address}="00:0c:6e:e4:2c:81", NAME="dmz"

This should fix it.

Thanks,
Kay
 
diff -Nru a/udev_lib.c b/udev_lib.c
--- a/udev_lib.c	2004-11-19 15:40:52 +01:00
+++ b/udev_lib.c	2004-11-19 15:40:52 +01:00
@@ -40,6 +40,7 @@
 		     const char *subsystem, const char* action)
 {
 	memset(udev, 0x00, sizeof(struct udevice));
+
 	if (devpath)
 		strfieldcpy(udev->devpath, devpath);
 	if (subsystem)
@@ -49,17 +50,13 @@
 
 	if (strcmp(udev->subsystem, "block") == 0)
 		udev->type = 'b';
-
-	if (strcmp(udev->subsystem, "net") == 0)
+	else if (strcmp(udev->subsystem, "net") == 0)
 		udev->type = 'n';
-
-	if (strncmp(udev->devpath, "/block/", 7) == 0)
+	else if (strncmp(udev->devpath, "/block/", 7) == 0)
 		udev->type = 'b';
-
-	if (strncmp(udev->devpath, "/class/net/", 11) == 0)
+	else if (strncmp(udev->devpath, "/class/net/", 11) == 0)
 		udev->type = 'n';
-
-	if (strncmp(udev->devpath, "/class/", 7) == 0)
+	else if (strncmp(udev->devpath, "/class/", 7) == 0)
 		udev->type = 'c';
 }