Duncan posted on Sat, 27 Jul 2013 22:14:02 +0000 as excerpted: > btrfs raid1 root here, was initr*less until I switched to btrfs which is > broken with direct-kernel-root-mount rootflags=device=whatever syntax. > > UUIDs are indeed userspace -- udev/systemd. However, if your initr* > includes udev, at least here, it "just works". > > I use root=LABEL=whatever here on the kernel commandline for root, and > LABEL=whatever for non-root in fstab, but as long as udev has the > directory in /dev/disk/*, mount should work with it, so > root=UUID=whatever at the kernel commandline should work, as should > UUID=whatever in fstab as the first field. I can add... * I use dracut as my initramfs generator, but with some of the default modules stripped in ordered to create a leaner initramfs. * It has a(n optional but obviously activated here) btrfs module that among other things, runs btrfs device scan before attempting to mount real-root. That's the critical bit that should be in your initramfs before attempting to mount a multi-device btrfs. With the btrfs executable and a call to btrfs device scan, mount, and udev creating the /dev/disk/by-*/ subdirs, an initramfs environment should really handle pretty much all the mount options available to you at a full-booted commandline. -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
