Closed solardiz closed 1 year ago
did not test the
off
count check code path in the hook.
However, made sure the expected version of it is compiled in, by temporarily inserting a #error
near it in:
#if P_OVL_OVERRIDE_SYNC_MODE == 2
return p_off == 3 * p_global_cnt_cookie;
#else
return p_off == 2 * p_global_cnt_cookie;
#endif
CI is all green both here and in my fork. @Adam-pi3 OK to merge?
This works for me too!
Done
Description
This simplifies the changes introduced in #217.
How Has This Been Tested?
Built and loaded for/into a kernel where
ovl_dentry_is_whiteout
is expected. Saw the expected log messages (unchanged from before, except that the wordfuncion
is dropped for consistency with other current LKRG messages) for the case of nooverlay
loaded yet. Did amodprobe overlay
, foundovl_dentry_is_whiteout
in/sys/kernel/debug/kprobes/list
. Unloaded and reloaded LKRG (withrmmod
andinsmod
), saw no messages about inability to hookovl_dentry_is_whiteout
that time, saw that it is hooked per/sys/kernel/debug/kprobes/list
.Did not test the logic beyond successful hooking of what's expected. That is, did not test the
off
count check code path in the hook.Meanwhile, 12 successful and 6 in progress CI checks in my fork.