[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190326003804.GK18020@tassilo.jf.intel.com>
Date: Mon, 25 Mar 2019 17:38:04 -0700
From: Andi Kleen <ak@...ux.intel.com>
To: Thomas Gleixner <tglx@...utronix.de>
Cc: "Chang S. Bae" <chang.seok.bae@...el.com>,
Ingo Molnar <mingo@...nel.org>,
Andy Lutomirski <luto@...nel.org>,
"H . Peter Anvin" <hpa@...or.com>,
Ravi Shankar <ravi.v.shankar@...el.com>,
LKML <linux-kernel@...r.kernel.org>,
Andrew Cooper <andrew.cooper3@...rix.com>, x86@...nel.org
Subject: Re: [RESEND PATCH v6 04/12] x86/fsgsbase/64: Enable FSGSBASE
instructions in the helper functions
> So on user space to kernel space transitions swapping in kernel GS should
> simply do:
> userGS = RDGSBASE()
> WRGSBASE(kernelGS)
This would also need to find kernelGS first, by doing RDPID and then
reading it from memory in the right index
(which might be a full cache miss if you're unlucky)
SWAPGS will be a lot faster, especially in these rare worst cases
because it has all its state inside the CPU.
-Andi
BTW you managed to only review after Chang went on a long vacation.
<rant>
I don't understand why it takes that long to review these changes
It's one of the largest performance improvements for the context
switch and the NMI in many years plus gives a new free register
to user space, but it only makes progress at a glacial pace.
The original patches for this were posted in 2016.
</rant>
Powered by blists - more mailing lists