x86/MCE: Serialize sysfs changes
authorSeunghun Han <[email protected]>
Tue, 6 Mar 2018 14:21:43 +0000 (15:21 +0100)
committerThomas Gleixner <[email protected]>
Thu, 8 Mar 2018 14:36:27 +0000 (15:36 +0100)
commitb3b7c4795ccab5be71f080774c45bbbcc75c2aaf
treee17563a8e944b6bc4c33bd587aa13383d6038c05
parentfa94d0c6e0f3431523f5701084d799c77c7d4a4f
x86/MCE: Serialize sysfs changes

The check_interval file in

  /sys/devices/system/machinecheck/machinecheck<cpu number>

directory is a global timer value for MCE polling. If it is changed by one
CPU, mce_restart() broadcasts the event to other CPUs to delete and restart
the MCE polling timer and __mcheck_cpu_init_timer() reinitializes the
mce_timer variable.

If more than one CPU writes a specific value to the check_interval file
concurrently, mce_timer is not protected from such concurrent accesses and
all kinds of explosions happen. Since only root can write to those sysfs
variables, the issue is not a big deal security-wise.

However, concurrent writes to these configuration variables is void of
reason so the proper thing to do is to serialize the access with a mutex.

Boris:

 - Make store_int_with_restart() use device_store_ulong() to filter out
   negative intervals
 - Limit min interval to 1 second
 - Correct locking
 - Massage commit message

Signed-off-by: Seunghun Han <[email protected]>
Signed-off-by: Borislav Petkov <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Tony Luck <[email protected]>
Cc: linux-edac <[email protected]>
Cc: [email protected]
Link: http://lkml.kernel.org/r/[email protected]
arch/x86/kernel/cpu/mcheck/mce.c