Re: [PATCH 2/5] x86, fpu: don't drop_fpu() in __restore_xstate_sig() if use_eager_fpu()

From: Linus Torvalds
Date: Sun Aug 24 2014 - 16:05:58 EST

I really dislike this one.

If I read it right, you now do *two* math_state_restore calls for each
FPU signal state restore. That's potentially quite expensive.

Also, you can actually end up with multiple threads pointing to the
same math state in init_task.thread.fpu.state, right? Why is that any
better than just having the save state temporarily contain garbage?

The other patches look sane, this one I really don't like. You may
have good reasons for it, but it's disgusting.


On Sun, Aug 24, 2014 at 12:47 PM, Oleg Nesterov <oleg@xxxxxxxxxx> wrote:
> __restore_xstate_sig() calls math_state_restore() with preemption
> enabled, not good. But this is minor, [...]
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at