Bryan Kadzban wrote: > Bruce Dubbs wrote: >> I'm trying to write a script to be run in Section 7.2 to initialize >> 70-persistent-net.rules. I'm using, as a base, the udev-182 rule >> 75-persistent-net-generator.rules. >> >> I have a problem in that this rule uses a variable SUBSYSTEMS. > > Uh oh, that's one thing I was afraid of. :-( Might be fixable, but it > might also be hard. > > SUBSYSTEMS is "the SUBSYSTEM of the matching parent device", but the > concept of a "parent device" is not exposed any obvious way in the sysfs > tree. The only way I know of to find it is to udevadm info > --attribute-walk. > > The SUBSYSTEM value for a given device directory is definitely the > basename of the target of the "subsystem" link; it's just that I don't > know how to find the parent devices. > > Looking at the libudev source (from -180), it appears that the algorithm > is simpler than I thought, though; it's as follows: > > 1. Start with the /sys path given, and resolve symlinks all the way down > 2. Remove one directory from the end of the path > 3. If the result of 2 is under /sys/devices, and the directory has a > "uevent" file present, then finish. > 4. If the result of 2 is under /sys/devices and the directory does not > have a "uevent" file present, then goto 2. > 5. If the result of 2 is not under /sys/devices, and the result is a > directory, then finish. (Would be hard for this to fail.) > 6. If the result of 2 is not under /sys/devices, and the result is not a > directory, then goto 2. > > The result after "finish" is then the next level in the tree. The > recursion stops when the code hits the /sys directory. > > So on my system, /sys/class/net/eth0 resolves to > /sys/devices/pci0000:00/0000:00:06.0/0000:05:05.0/net/eth0. That has a > uevent file. The first parent is the "net" directory, which does not > have one, so the algorithm continues to the "0000:05:05.0" directory. > That does have a uevent file, so that's next in line. The 0000:00:06.0 > directory is next, then the pci0000:00 directory. The "devices" > directory doesn't have a uevent file, so the next loop iteration is the > last one (hits /sys), which doesn't find a parent for the pci0000:00 > directory.
What I'm using right now is: # Ignore Xen virtual interfaces if [ -e /proc/xen ]; then msg="The rules file should not be created in the Xen environment" usage fi I'm not sure if that is right or not. Someone with Xen needs to verify. I also skipped the other SUBSYSTEMS values. They are just contributing to a comment and I don't think that's required. I did find NIC=/sys/class/net/eth0 $ basename $(readlink -f $NIC/device/subsystem) pci but I'm not sure how robust that is. I'm going to post a new tarball in a few minutes after I finish some testing. Take a look at that. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
