[PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[Tom Lendacky]

AMD processors are not subject to the types of attacks that the kernel
page table isolation feature protects against.  The AMD microarchitecture
does not allow memory references, including speculative references, that
access higher privileged data when running in a lesser privileged mode
when that access would result in a page fault.

Disable page table isolation by default on AMD processors by not setting
the X86_BUG_CPU_INSECURE feature, which controls whether X86_FEATURE_PTI
is set.

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
---
 arch/x86/kernel/cpu/common.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index c47de4e..7d9e3b0 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -923,8 +923,8 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c)
 
 	setup_force_cpu_cap(X86_FEATURE_ALWAYS);
 
-	/* Assume for now that ALL x86 CPUs are insecure */
-	setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
+	if (c->x86_vendor != X86_VENDOR_AMD)
+		setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
 
 	fpu__init_system(c);
 

Re: [PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[Dave Hansen]

On 12/26/2017 09:43 PM, Tom Lendacky wrote:
> --- a/arch/x86/kernel/cpu/common.c
> +++ b/arch/x86/kernel/cpu/common.c
> @@ -923,8 +923,8 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c)
>  
>  	setup_force_cpu_cap(X86_FEATURE_ALWAYS);
>  
> -	/* Assume for now that ALL x86 CPUs are insecure */
> -	setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
> +	if (c->x86_vendor != X86_VENDOR_AMD)
> +		setup_force_cpu_bug(X86_BUG_CPU_INSECURE);

Does this disable it in a way that it can be turned back on via the
kernel command-line?

This is a rather wide class of issues and I would rather not just
hard-code it in a way that we say one vendor has never and will never be
affected.

Re: [PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[Tom Lendacky]

On 12/27/2017 2:48 AM, Dave Hansen wrote:
> On 12/26/2017 09:43 PM, Tom Lendacky wrote:
>> --- a/arch/x86/kernel/cpu/common.c
>> +++ b/arch/x86/kernel/cpu/common.c
>> @@ -923,8 +923,8 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c)
>>  
>>  	setup_force_cpu_cap(X86_FEATURE_ALWAYS);
>>  
>> -	/* Assume for now that ALL x86 CPUs are insecure */
>> -	setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
>> +	if (c->x86_vendor != X86_VENDOR_AMD)
>> +		setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
> 
> Does this disable it in a way that it can be turned back on via the
> kernel command-line?
> 

Yes, specifying pti=on on the command line will turn kernel page table
isolation on regardless of this setting.

Thanks,
Tom

> This is a rather wide class of issues and I would rather not just
> hard-code it in a way that we say one vendor has never and will never be
> affected.
> 

Re: [PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[Borislav Petkov]

On Tue, Dec 26, 2017 at 11:43:54PM -0600, Tom Lendacky wrote:
> AMD processors are not subject to the types of attacks that the kernel
> page table isolation feature protects against.  The AMD microarchitecture
> does not allow memory references, including speculative references, that
> access higher privileged data when running in a lesser privileged mode
> when that access would result in a page fault.
> 
> Disable page table isolation by default on AMD processors by not setting
> the X86_BUG_CPU_INSECURE feature, which controls whether X86_FEATURE_PTI
> is set.
> 
> Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
> ---
>  arch/x86/kernel/cpu/common.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
> index c47de4e..7d9e3b0 100644
> --- a/arch/x86/kernel/cpu/common.c
> +++ b/arch/x86/kernel/cpu/common.c
> @@ -923,8 +923,8 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c)
>  
>  	setup_force_cpu_cap(X86_FEATURE_ALWAYS);
>  
> -	/* Assume for now that ALL x86 CPUs are insecure */
> -	setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
> +	if (c->x86_vendor != X86_VENDOR_AMD)
> +		setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
>  
>  	fpu__init_system(c);

Reviewed-by: Borislav Petkov <bp@suse.de>

-- 
Regards/Gruss,
    Boris.

SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg)
-- 

[tip:x86/pti] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[tip-bot for Tom Lendacky]

Commit-ID:  694d99d40972f12e59a3696effee8a376b79d7c8
Gitweb:     https://git.kernel.org/tip/694d99d40972f12e59a3696effee8a376b79d7c8
Author:     Tom Lendacky <thomas.lendacky@amd.com>
AuthorDate: Tue, 26 Dec 2017 23:43:54 -0600
Committer:  Thomas Gleixner <tglx@linutronix.de>
CommitDate: Wed, 3 Jan 2018 15:57:59 +0100

x86/cpu, x86/pti: Do not enable PTI on AMD processors

AMD processors are not subject to the types of attacks that the kernel
page table isolation feature protects against.  The AMD microarchitecture
does not allow memory references, including speculative references, that
access higher privileged data when running in a lesser privileged mode
when that access would result in a page fault.

Disable page table isolation by default on AMD processors by not setting
the X86_BUG_CPU_INSECURE feature, which controls whether X86_FEATURE_PTI
is set.

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20171227054354.20369.94587.stgit@tlendack-t1.amdoffice.net

---
 arch/x86/kernel/cpu/common.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index f2a94df..b1be494 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -899,8 +899,8 @@ static void __init early_identify_cpu(struct cpuinfo_x86 *c)
 
 	setup_force_cpu_cap(X86_FEATURE_ALWAYS);
 
-	/* Assume for now that ALL x86 CPUs are insecure */
-	setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
+	if (c->x86_vendor != X86_VENDOR_AMD)
+		setup_force_cpu_bug(X86_BUG_CPU_INSECURE);
 
 	fpu__init_system(c);
 

Re: [PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 867 bytes --]

On Tue 2017-12-26 23:43:54, Tom Lendacky wrote:
> AMD processors are not subject to the types of attacks that the kernel
> page table isolation feature protects against.  The AMD microarchitecture
> does not allow memory references, including speculative references, that
> access higher privileged data when running in a lesser privileged mode
> when that access would result in a page fault.
> 
> Disable page table isolation by default on AMD processors by not setting
> the X86_BUG_CPU_INSECURE feature, which controls whether X86_FEATURE_PTI
> is set.

PTI was originally meant to protect KASLR from memory leaks, before
Spectre was public. I guess that's still valid use on AMD cpus?
								Pavel
								
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[Brian Gerst]

On Mon, Feb 12, 2018 at 10:26 AM, Pavel Machek <pavel@ucw.cz> wrote:
> On Tue 2017-12-26 23:43:54, Tom Lendacky wrote:
>> AMD processors are not subject to the types of attacks that the kernel
>> page table isolation feature protects against.  The AMD microarchitecture
>> does not allow memory references, including speculative references, that
>> access higher privileged data when running in a lesser privileged mode
>> when that access would result in a page fault.
>>
>> Disable page table isolation by default on AMD processors by not setting
>> the X86_BUG_CPU_INSECURE feature, which controls whether X86_FEATURE_PTI
>> is set.
>
> PTI was originally meant to protect KASLR from memory leaks, before
> Spectre was public. I guess that's still valid use on AMD cpus?
>                                                                 Pavel

KASLR leaks are a much lower threat than Meltdown.  Given that no AMD
processor supports PCID, enabling PTI has a much more significant
performance impact for a much smaller benefit.  For the paranoid user
they still have the option to enable PTI at boot, but it should not be
on by default.

--
Brian Gerst

Re: [PATCH] x86/cpu, x86/pti: Do not enable PTI on AMD processors

[Thomas Gleixner]

On Mon, 12 Feb 2018, Pavel Machek wrote:

> On Tue 2017-12-26 23:43:54, Tom Lendacky wrote:
> > AMD processors are not subject to the types of attacks that the kernel
> > page table isolation feature protects against.  The AMD microarchitecture
> > does not allow memory references, including speculative references, that
> > access higher privileged data when running in a lesser privileged mode
> > when that access would result in a page fault.
> > 
> > Disable page table isolation by default on AMD processors by not setting
> > the X86_BUG_CPU_INSECURE feature, which controls whether X86_FEATURE_PTI
> > is set.
> 
> PTI was originally meant to protect KASLR from memory leaks, before
> Spectre was public. I guess that's still valid use on AMD cpus?

The KASLR attacks against which PTI protects are not based on a memory
leak. The KASLR attacks are revealing the kernel virtual address space w/o
revealing any data.

Quite some of those attacks can be mitigated via PTI, but only some of the
attacks work on AMD CPUs. The bulk (and easy to conduct) attacks do not
work work on AMD CPUs due to the same reason why Meltdown does not work.

Thanks,

	tglx