linux-kernel - Re: [tip:x86/pti] x86/pti: Make sure the user/kernel PTEs match

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Tue, 13 Feb 2018 16:59:23 +0100 (CET)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Dave Hansen <dave.hansen@...el.com>
cc:     bp@...en8.de, mingo@...nel.org, linux-kernel@...r.kernel.org,
        mroos@...ux.ee, hpa@...or.com, thomas.lendacky@....com,
        linux-tip-commits@...r.kernel.org
Subject: Re: [tip:x86/pti] x86/pti: Make sure the user/kernel PTEs match

On Fri, 9 Feb 2018, Dave Hansen wrote:
> On 01/03/2018 08:22 AM, tip-bot for Thomas Gleixner wrote:
> > --- a/arch/x86/mm/pti.c
> > +++ b/arch/x86/mm/pti.c
> > @@ -367,7 +367,8 @@ static void __init pti_setup_espfix64(void)
> >  static void __init pti_clone_entry_text(void)
> >  {
> >  	pti_clone_pmds((unsigned long) __entry_text_start,
> > -			(unsigned long) __irqentry_text_end, _PAGE_RW);
> > +			(unsigned long) __irqentry_text_end,
> > +		       _PAGE_RW | _PAGE_GLOBAL);
> >  }
> 
> We appear to have lost all of our global bits between this and the
> cpu_entry_area itself.  To make K8 happy, shouldn't we just *set*
> _PAGE_GLOBAL universally instead of clearing it universally?
> 
> Also, don't we want _PAGE_GLOBAL in the cpu_entry_area.c pte setting?  I
> think I had this for all the shared user/kernel areas near the end of
> the KAISER era, but we appear to have totally nuked it for PTI.

Yes, we did due to the usage of supported_pte_flags vs. the hard coded
compile time thing. It was on my todo list to flag the shared areas global,
but I somehow nuked that item in a vein attempt to shrink that list :)
We surely should revisit that.

Thanks,

	tglx