Closed stephensmalley closed 4 years ago
Related to #51.
Adding @dburgener.
Partly fixed via 02a52c5c8c3b8cbad0f12009cde9f36dbefb6972; an error while updating selinuxfs will now leave the system running with the old policy. selinuxfs itself may still be in an inconsistent state; this will be addressed via #51.
Presently SELinux policy reload is not atomic wrt updating selinuxfs, so a failure while updating selinuxfs (e.g. while creating the new booleans or class subdirectories) will leave the system in an inconsistent state. Fixing this requires refactoring policy load as per https://lore.kernel.org/selinux/20181002155810.GP32577@ZenIV.linux.org.uk/