The existence of /sys/firmware/efi/efivars only indicates that we've
booted the kernel in EFI mode. It does not guarantee that the same
kernel has support for efivarfs. Since this isn't critical, don't fail
the whole api mount.
This is potentially evil, but we're using hand crafted input to eval
which should be safe. Use this to bind mount the chroot dir if it isn't
already a mount point.
This was brought up when a user tried to arch-chroot into the same
root twice from different terminals. The second chroot should fail, but
not tear down the tree that it doesn't own.
We're kind enough to provide DM names for non-swap devices, but simply
use the /dev/dm-X nodes for swap, since that's what /proc/swaps provides
us with. Resolve the DM name out of sysfs and use this instead.
Fixes FS#32609.
This mainly fixes the case where there's devices mounted under the
chroot directory. Creating a bind mount on top of this will obscure
these other mounts.
The %b formatter is too generic for us to use, as a sequence such as
'\b' is invalid for hex escaping, but valid for %b. Port over the logic
from util-linux's mangle mini-lib and use this instead. This supports
both 3 digits hex as well as 3 digit octal escapes, and will ignore the
more generic ascii based single character escapes. Output is always in
octal format to comply with fstab(5).
* the Makefile uses m4 to build the common lib into the scripts
* it includes install, uninstall and clean targets
* the dist target creates a source tar from a tag
Signed-off-by: Pierre Schmitz <pierre@archlinux.de>
