On Tue, Sep 02, 2025 at 10:57:57AM -0400, Steven Rostedt wrote:

And then have:

       quiet_rcu_read_lock_check();
       head = rhltable_lookup(&fprobe_ip_table, &func, fprobe_rht_params);
       quiet_rcu_read_unlock_check();

The thing is that rhl_for_each_entry_rcu which is called right after
your unlock above should have created the same warning as
rhltable_lookup.  The fact that it doesn't appears to be a bug:
it's using rcu_dereference_raw and I don't see why that's safe
at all.

Cheers,
-- 
Email: Herbert Xu [off-list ref]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt