Revert 337f13046ff0 ("futex: Allow FUTEX_CLOCK_REALTIME with FUTEX_WAIT op")
authorThomas Gleixner <tglx@linutronix.de>
Thu, 22 Apr 2021 19:44:18 +0000 (21:44 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 22 May 2021 08:40:19 +0000 (10:40 +0200)
commita1eb878265b8ad200037fabe467d12d712445e1d
tree7bdc98ed28c622bab364d625a38aa34672d5316c
parent6fa27e51e4d32a55f0d83e8c4bc1799528ac674b
Revert 337f13046ff0 ("futex: Allow FUTEX_CLOCK_REALTIME with FUTEX_WAIT op")

commit 4fbf5d6837bf81fd7a27d771358f4ee6c4f243f8 upstream.

The FUTEX_WAIT operand has historically a relative timeout which means that
the clock id is irrelevant as relative timeouts on CLOCK_REALTIME are not
subject to wall clock changes and therefore are mapped by the kernel to
CLOCK_MONOTONIC for simplicity.

If a caller would set FUTEX_CLOCK_REALTIME for FUTEX_WAIT the timeout is
still treated relative vs. CLOCK_MONOTONIC and then the wait arms that
timeout based on CLOCK_REALTIME which is broken and obviously has never
been used or even tested.

Reject any attempt to use FUTEX_CLOCK_REALTIME with FUTEX_WAIT again.

The desired functionality can be achieved with FUTEX_WAIT_BITSET and a
FUTEX_BITSET_MATCH_ANY argument.

Fixes: 337f13046ff0 ("futex: Allow FUTEX_CLOCK_REALTIME with FUTEX_WAIT op")
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20210422194704.834797921@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
kernel/futex.c