linux-kernel - One potential issue with concurrent execution of RCU callbacks...

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 8 Dec 2020 06:58:10 -0800
From:   "Paul E. McKenney" <paulmck@...nel.org>
To:     frederic@...nel.org
Cc:     boqun.feng@...il.com, rcu@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: One potential issue with concurrent execution of RCU callbacks...

Hello, Frederic,

Boqun just asked if RCU callbacks ran in BH-disabled context to avoid
concurrent execution of the same callback.  Of course, this raises the
question of whether a self-posting callback can have two instances of
itself running concurrently while a CPU is in the process of transitioning
between softirq and rcuo invocation of callbacks.

I believe that the answer is "no" because BH-disabled context is
an implicit RCU read-side critical section.  Therefore, the initial
invocation of the RCU callback must complete in order for a new grace
period to complete, and a new grace period must complete before the
second invocation of that same callback to start.

Does that make sense, or am I missing something?

							Thanx, Paul