x86, fpu: Fix math_state_restore() race with kernel_fpu_begin()
authorOleg Nesterov <oleg@redhat.com>
Thu, 15 Jan 2015 19:20:28 +0000 (20:20 +0100)
committerThomas Gleixner <tglx@linutronix.de>
Tue, 20 Jan 2015 12:53:07 +0000 (13:53 +0100)
commit7575637ab293861a799f3bbafe0d8c597389f4e9
tree52cc8c7a21eecaa53bbe3222864bcd3d8e482851
parent33a3ebdc077fd85f1bf4d4586eea579b297461ae
x86, fpu: Fix math_state_restore() race with kernel_fpu_begin()

math_state_restore() can race with kernel_fpu_begin() if irq comes
right after __thread_fpu_begin(), __save_init_fpu() will overwrite
fpu->state we are going to restore.

Add 2 simple helpers, kernel_fpu_disable() and kernel_fpu_enable()
which simply set/clear in_kernel_fpu, and change math_state_restore()
to exclude kernel_fpu_begin() in between.

Alternatively we could use local_irq_save/restore, but probably these
new helpers can have more users.

Perhaps they should disable/enable preemption themselves, in this case
we can remove preempt_disable() in __restore_xstate_sig().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: matt.fleming@intel.com
Cc: bp@suse.de
Cc: pbonzini@redhat.com
Cc: luto@amacapital.net
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Link: http://lkml.kernel.org/r/20150115192028.GD27332@redhat.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
arch/x86/include/asm/i387.h
arch/x86/kernel/i387.c
arch/x86/kernel/traps.c