This avoids that a following sync syscall in in pm_suspend()
(called when we echo mem > /sys/power/state) hangs forever
in at least one case (on my machine).
I guess that one case is the following: if tracking access time
is enabled on an ext4 file system, then executing chroot in
arch-luks-suspend will require I/O in the sync syscall in
pm_suspend().
Anyway, it cannot be wrong to do the sync as late as possible.
This avoids that a following sync syscall in in pm_suspend() (called when we echo mem > /sys/power/state) hangs forever in at least one case (on my machine).
I guess that one case is the following: if tracking access time is enabled on an ext4 file system, then executing chroot in arch-luks-suspend will require I/O in the sync syscall in pm_suspend().
Anyway, it cannot be wrong to do the sync as late as possible.