[PATCH v3] x86/bugs: Default retbleed to =stuff when retpoline is enabled

[PATCH v3] x86/bugs: Default retbleed to =stuff when retpoline is enabled

[Pawan Gupta]

On Intel systems when retpoline mitigation is enabled for spectre-v2,
retbleed=auto does not enable RSB stuffing. This may make the system
vulnerable to retbleed. Retpoline is not the default mitigation when
IBRS is present, but in virtualized cases a VMM can hide IBRS from
guests, resulting in guest deploying retpoline by default. Even if IBRS
is enumerated, a user can still select spectre_v2=retpoline.

As with other mitigations, mitigate retbleed by default. On Intel
systems when retpoline is enabled, and retbleed mitigation is set to
auto, enable Call Depth Tracking and RSB stuffing i.e. retbleed=stuff
mitigation. For AMD/Hygon auto mode already selects the appropriate
mitigation.

Reported-by: Alyssa Milburn <alyssa.milburn@intel.com>
Cc: stable@kernel.org
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
---
v3:
- Rebased to v6.9-rc2

v2: https://lore.kernel.org/r/20240212-retbleed-auto-stuff-v2-1-89401649341a@linux.intel.com
- Mitigate retbleed by default for spectre_v2=retpoline. (Josh)
- Add the missing ',' in the comment. (Josh)
- Rebased to v6.8-rc4

v1: https://lore.kernel.org/r/20240208-retbleed-auto-stuff-v1-1-6f12e513868f@linux.intel.com
---
 arch/x86/kernel/cpu/bugs.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
index e7ba936d798b..69d8ce58f244 100644
--- a/arch/x86/kernel/cpu/bugs.c
+++ b/arch/x86/kernel/cpu/bugs.c
@@ -1092,11 +1092,17 @@ static void __init retbleed_select_mitigation(void)
 			else if (IS_ENABLED(CONFIG_MITIGATION_IBPB_ENTRY) &&
 				 boot_cpu_has(X86_FEATURE_IBPB))
 				retbleed_mitigation = RETBLEED_MITIGATION_IBPB;
+		} else if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL &&
+			   spectre_v2_enabled == SPECTRE_V2_RETPOLINE) {
+			if (IS_ENABLED(CONFIG_CALL_DEPTH_TRACKING))
+				retbleed_mitigation = RETBLEED_MITIGATION_STUFF;
+			else
+				pr_err("WARNING: Retpoline enabled, but kernel not compiled with CALL_DEPTH_TRACKING.\n");
 		}
 
 		/*
-		 * The Intel mitigation (IBRS or eIBRS) was already selected in
-		 * spectre_v2_select_mitigation().  'retbleed_mitigation' will
+		 * If Intel mitigation (IBRS or eIBRS) was already selected in
+		 * spectre_v2_select_mitigation(),  'retbleed_mitigation' will
 		 * be set accordingly below.
 		 */
 

---
base-commit: 39cd87c4eb2b893354f3b850f916353f2658ae6f
change-id: 20240208-retbleed-auto-stuff-53e0fa91305e

Re: [PATCH v3] x86/bugs: Default retbleed to =stuff when retpoline is enabled

[Nikolay Borisov]

On 3.04.24 г. 8:54 ч., Pawan Gupta wrote:
> On Intel systems when retpoline mitigation is enabled for spectre-v2,
> retbleed=auto does not enable RSB stuffing. This may make the system
> vulnerable to retbleed. Retpoline is not the default mitigation when
> IBRS is present, but in virtualized cases a VMM can hide IBRS from
> guests, resulting in guest deploying retpoline by default. Even if IBRS
> is enumerated, a user can still select spectre_v2=retpoline.
> 
> As with other mitigations, mitigate retbleed by default. On Intel
> systems when retpoline is enabled, and retbleed mitigation is set to
> auto, enable Call Depth Tracking and RSB stuffing i.e. retbleed=stuff
> mitigation. For AMD/Hygon auto mode already selects the appropriate
> mitigation.
> 
> Reported-by: Alyssa Milburn <alyssa.milburn@intel.com>
> Cc: stable@kernel.org
> Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>

Reviewed-by: Nikolay Borisov <nik.borisov@suse.com>
> ---
> v3:
> - Rebased to v6.9-rc2
> 
> v2: https://lore.kernel.org/r/20240212-retbleed-auto-stuff-v2-1-89401649341a@linux.intel.com
> - Mitigate retbleed by default for spectre_v2=retpoline. (Josh)
> - Add the missing ',' in the comment. (Josh)
> - Rebased to v6.8-rc4
> 
> v1: https://lore.kernel.org/r/20240208-retbleed-auto-stuff-v1-1-6f12e513868f@linux.intel.com
> ---
>   arch/x86/kernel/cpu/bugs.c | 10 ++++++++--
>   1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
> index e7ba936d798b..69d8ce58f244 100644
> --- a/arch/x86/kernel/cpu/bugs.c
> +++ b/arch/x86/kernel/cpu/bugs.c
> @@ -1092,11 +1092,17 @@ static void __init retbleed_select_mitigation(void)
>   			else if (IS_ENABLED(CONFIG_MITIGATION_IBPB_ENTRY) &&
>   				 boot_cpu_has(X86_FEATURE_IBPB))
>   				retbleed_mitigation = RETBLEED_MITIGATION_IBPB;
> +		} else if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL &&
> +			   spectre_v2_enabled == SPECTRE_V2_RETPOLINE) {
> +			if (IS_ENABLED(CONFIG_CALL_DEPTH_TRACKING))
> +				retbleed_mitigation = RETBLEED_MITIGATION_STUFF;
> +			else
> +				pr_err("WARNING: Retpoline enabled, but kernel not compiled with CALL_DEPTH_TRACKING.\n");
>   		}
>   
>   		/*
> -		 * The Intel mitigation (IBRS or eIBRS) was already selected in
> -		 * spectre_v2_select_mitigation().  'retbleed_mitigation' will
> +		 * If Intel mitigation (IBRS or eIBRS) was already selected in
> +		 * spectre_v2_select_mitigation(),  'retbleed_mitigation' will
>   		 * be set accordingly below.
>   		 */
>   

nit: I think those outer 'ifs' might be a bit easier to grok if
they are turned into a switch:


26 do_cmd_auto:
     25         case RETBLEED_CMD_AUTO:
     24                 switch(boot_cpu_data.x86_vendor) {
     23                 case X86_VENDOR_AMD:
     22                 case X86_VENDOR_HYGON:
     21                         if (IS_ENABLED(CONFIG_MITIGATION_UNRET_ENTRY))
     20                                 retbleed_mitigation = RETBLEED_MITIGATION_UNRET;
     19                         else if (IS_ENABLED(CONFIG_MITIGATION_IBPB_ENTRY) &&
     18                                  boot_cpu_has(X86_FEATURE_IBPB))
     17                                 retbleed_mitigation = RETBLEED_MITIGATION_IBPB;
     16                         break;
     15                 case X86_VENDOR_INTEL:
     14                         if (spectre_v2_enabled == SPECTRE_V2_RETPOLINE) {
     13                                 if (IS_ENABLED(CONFIG_CALL_DEPTH_TRACKING))
     12                                         retbleed_mitigation = RETBLEED_MITIGATION_STUFF;
     11                                 else
     10                                         pr_err("WARNING: Retpoline enabled, but kernel not compiled with CALL_DEPTH_TRACKING.\n");
      9                         }
      8
      7                         /*
      6                          * The Intel mitigation (IBRS or eIBRS) was already selected in
      5                          * spectre_v2_select_mitigation().  'retbleed_mitigation' will
      4                          * be set accordingly below.
      3                          */
      2                 default:
      1                         break;
   1112                 }





> 
> ---
> base-commit: 39cd87c4eb2b893354f3b850f916353f2658ae6f
> change-id: 20240208-retbleed-auto-stuff-53e0fa91305e
> 
> 

Re: [PATCH v3] x86/bugs: Default retbleed to =stuff when retpoline is enabled

[Pawan Gupta]

On Wed, Apr 03, 2024 at 05:32:40PM +0300, Nikolay Borisov wrote:
> 
> 
> On 3.04.24 г. 8:54 ч., Pawan Gupta wrote:
> > On Intel systems when retpoline mitigation is enabled for spectre-v2,
> > retbleed=auto does not enable RSB stuffing. This may make the system
> > vulnerable to retbleed. Retpoline is not the default mitigation when
> > IBRS is present, but in virtualized cases a VMM can hide IBRS from
> > guests, resulting in guest deploying retpoline by default. Even if IBRS
> > is enumerated, a user can still select spectre_v2=retpoline.
> > 
> > As with other mitigations, mitigate retbleed by default. On Intel
> > systems when retpoline is enabled, and retbleed mitigation is set to
> > auto, enable Call Depth Tracking and RSB stuffing i.e. retbleed=stuff
> > mitigation. For AMD/Hygon auto mode already selects the appropriate
> > mitigation.
> > 
> > Reported-by: Alyssa Milburn <alyssa.milburn@intel.com>
> > Cc: stable@kernel.org
> > Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
> 
> Reviewed-by: Nikolay Borisov <nik.borisov@suse.com>

Thanks.