x86/debug: Prevent data breakpoints on __per_cpu_offset
authorLai Jiangshan <laijs@linux.alibaba.com>
Thu, 4 Feb 2021 15:27:06 +0000 (23:27 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 10 Feb 2021 08:29:21 +0000 (09:29 +0100)
commitb796770c6db3dabe0daa7d7d1f04cc6afd53c569
treedd02dd64f9e16c28822341bdfdbf663caf15821c
parentc000dcfb3aede952db92086053d03be3da30487e
x86/debug: Prevent data breakpoints on __per_cpu_offset

commit c4bed4b96918ff1d062ee81fdae4d207da4fa9b0 upstream.

When FSGSBASE is enabled, paranoid_entry() fetches the per-CPU GSBASE value
via __per_cpu_offset or pcpu_unit_offsets.

When a data breakpoint is set on __per_cpu_offset[cpu] (read-write
operation), the specific CPU will be stuck in an infinite #DB loop.

RCU will try to send an NMI to the specific CPU, but it is not working
either since NMI also relies on paranoid_entry(). Which means it's
undebuggable.

Fixes: eaad981291ee3("x86/entry/64: Introduce the FIND_PERCPU_BASE macro")
Signed-off-by: Lai Jiangshan <laijs@linux.alibaba.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20210204152708.21308-1-jiangshanlai@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/x86/kernel/hw_breakpoint.c