Re: [PATCH 0/3] Use global pages with PTI

[Dave Hansen]

On 02/15/2018 09:47 AM, Linus Torvalds wrote:
> On Thu, Feb 15, 2018 at 5:20 AM, Dave Hansen
> <dave.hansen@xxxxxxxxxxxxxxx> wrote:
>>
>> During the switch over to PTI, we seem to have lost our ability to have
>> GLOBAL mappings.
> 
> Oops. Odd, I have this distinct memory of somebody even _testing_ the
> global bit performance when I pointed out that we shouldn't just make
> the bit go away entirely.
> 
> [ goes back and looks at archives ]
> 
> Oh, that was in fact you who did that performance test.
...
> Did you perhaps re-run any benchmark numbers just to verify? Because
> it's always good to back up patches that should improve performance
> with actual numbers..

Nope, haven't done it yet, but I will.

I wanted to double-check that there was not a reason for doing the
global disabling other than the K8 TLB mismatch issues that Thomas fixed
a few weeks ago:

> commit 52994c256df36fda9a715697431cba9daecb6b11
> Author: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Date:   Wed Jan 3 15:57:59 2018 +0100
> 
>     x86/pti: Make sure the user/kernel PTEs match
>     
>     Meelis reported that his K8 Athlon64 emits MCE warnings when PTI is
>     enabled:
>     
>     [Hardware Error]: Error Addr: 0x0000ffff81e000e0
>     [Hardware Error]: MC1 Error: L1 TLB multimatch.
>     [Hardware Error]: cache level: L1, tx: INSN