From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Message-ID: <52ED562B.60703@zytor.com> Date: Sat, 01 Feb 2014 12:16:43 -0800 From: "H. Peter Anvin" MIME-Version: 1.0 To: Linus Torvalds CC: Suresh Siddha , Nate Eldredge , Thomas Gleixner , Ingo Molnar , the arch/x86 maintainers , stable , Linux Kernel Mailing List , Maarten Baert , Jan Kara , George Spelvin , Pekka Riikonen Subject: Re: [PATCH] Make math_state_restore() save and restore the interrupt flag References: <52ED4C96.6020703@zytor.com> In-Reply-To: Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: On 02/01/2014 11:46 AM, Linus Torvalds wrote: > > .. which *does* actually bring up something that might work, and might > be a good idea: remove the "restore math state or set TS" from the > normal kernel paths *entirely*, and move it to the "return to user > space" phase. > The task switch code would have to be aware in case of preemption, though. The nice thing about *that* is that we could probably completely drop the notion that we can't preempt an FPU-using kernel task. -hpa