linux-kernel - Re: [PATCH v2] Linux VM workaround for Knights Landing A/D leak

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

Date:	Tue, 14 Jun 2016 13:20:06 -0700
From:	"H. Peter Anvin" <hpa@...or.com>
To:	Borislav Petkov <bp@...en8.de>,
	Dave Hansen <dave.hansen@...ux.intel.com>
Cc:	One Thousand Gnomes <gnomes@...rguk.ukuu.org.uk>,
	Lukasz Anaczkowski <lukasz.anaczkowski@...el.com>,
	linux-kernel@...r.kernel.org, linux-mm@...ck.org,
	tglx@...utronix.de, mingo@...hat.com, ak@...ux.intel.com,
	kirill.shutemov@...ux.intel.com, mhocko@...e.com,
	akpm@...ux-foundation.org, harish.srinivasappa@...el.com,
	lukasz.odzioba@...el.com, grzegorz.andrejczuk@...el.com,
	lukasz.daniluk@...el.com
Subject: Re: [PATCH v2] Linux VM workaround for Knights Landing A/D leak

On 06/14/16 12:19, Borislav Petkov wrote:
> On Tue, Jun 14, 2016 at 11:54:24AM -0700, Dave Hansen wrote:
>> Lukasz, Borislav suggested using static_cpu_has_bug(), which will do the
>> alternatives patching.  It's definitely the right thing to use here.
> 
> Yeah, either that or do an
> 
> alternative_call(null_func, fix_pte_peak, X86_BUG_PTE_LEAK, ...)
> 
> or so and you'll need a dummy function to call on !X86_BUG_PTE_LEAK
> CPUs.
> 
> The static_cpu_has_bug() thing should be most likely a penalty
> of a single JMP (I have to look at the asm) but then since the
> callers are inlined, you'll have to patch all those places where
> *ptep_get_and_clear() get inlined.
> 
> Shouldn't be a big deal still but...
> 
> "debug-alternative" and a kvm guest should help you there to get a quick
> idea.
> 

static_cpu_has_bug() should turn into 5-byte NOP in the common (bugless)
case.

	-hpa