Skip to contentSkip to conversation

x86/fpu: Disable shstk if no CET_USER state

PATCH Superseded David Kaplan 2 participants 8 messages Apr 3 – Apr 7 x86

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/fpu: Disable shstk if no CET_USER state

Some hypervisors (including QEMU 10.1.5) may report CET_SS support in CPUID Fn7 but fail to report that CET_USER state is supported in supervisor xstate. Linux relies on XSAVES/XRSTORS to swap CET state during context switch and assumes it is supported when CET_SS is present.

As a result, if a user process is run with shadow stacks enabled and then is switched away from, the system may crash because the new process may be incorrectly run with shadow stacks enabled.

Detect this broken configuration and disable user shadow stacks unless CET_USER is supported in xstate.

---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
arch/x86/kernel/fpu/xstate.c +11 -0
index 76153dfb58c9..188323442b4d 100644
@@ -855,6 +855,17 @@ void __init fpu__init_system_xstate(unsigned int legacy_size)
 		goto out_disable;
 	}
 
+	if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+	    !(fpu_kernel_cfg.max_features & XFEATURE_MASK_CET_USER)) {
+		/*
+		 * The kernel relies on XSAVES/XRSTORS to context switch shadow
+		 * stack state.  If this isn't present, disable user shadow
+		 * stacks.
+		 */
+		pr_err("x86/fpu: CET_USER not supported in xstate when CET is supported.  Disabling shadow stacks.\n");
+		setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);
+	}
+
 	fpu_kernel_cfg.independent_features = fpu_kernel_cfg.max_features &
 					      XFEATURE_MASK_INDEPENDENT;
 

base-commit: d998c62f267213aeb815cf654908608eb7c00db2

Signature · 1 line
2.53.0
#1 lore raw
Quoted 6 lines · David Kaplan

Some hypervisors (including QEMU 10.1.5) may report CET_SS support in CPUID Fn7 but fail to report that CET_USER state is supported in supervisor xstate. Linux relies on XSAVES/XRSTORS to swap CET state during context switch and assumes it is supported when CET_SS is present.

Quoted 6 lines

As a result, if a user process is run with shadow stacks enabled and then is switched away from, the system may crash because the new process may be incorrectly run with shadow stacks enabled.

Detect this broken configuration and disable user shadow stacks unless CET_USER is supported in xstate.

It's not actually broken though, is it? Just "odd". AFAICT, neither the SDM nor the APM _requires_ CET_{U,S} to be supported in XSS if shadow stacks are suppported.

Quoted 12 lines
---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
arch/x86/kernel/fpu/xstate.c +0 -0
index 76153dfb58c9..188323442b4d 100644
@@ -855,6 +855,17 @@ void __init fpu__init_system_xstate(unsigned int legacy_size)
 		goto out_disable;
 	}

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size) · line 866

Quoted 10 lines
arch/x86/kernel/fpu/xstate.c +9 -0
 
+	if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+	    !(fpu_kernel_cfg.max_features & XFEATURE_MASK_CET_USER)) {
+		/*
+		 * The kernel relies on XSAVES/XRSTORS to context switch shadow
+		 * stack state.  If this isn't present, disable user shadow
+		 * stacks.
+		 */
+		pr_err("x86/fpu: CET_USER not supported in xstate when CET is supported.  Disabling shadow stacks.\n");
+		setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);

Doesn't this apply to IBT as well? This code is also misplaced, as it needs to live after at least this code:

	if (!cpu_feature_enabled(X86_FEATURE_XSAVES))
		fpu_kernel_cfg.max_features &= XFEATURE_MASK_USER_SUPPORTED;
	else
		fpu_kernel_cfg.max_features &= XFEATURE_MASK_USER_SUPPORTED |
					XFEATURE_MASK_SUPERVISOR_SUPPORTED;

and should probably play nice with the "out_disable" path too.

All in all, setup_cet() seems like a much better fit, but unfortunately that runs before fpu__init_system() :-(

Quoted 10 lines
arch/x86/kernel/fpu/xstate.c +2 -0
+	}
+
 	fpu_kernel_cfg.independent_features = fpu_kernel_cfg.max_features &
 					      XFEATURE_MASK_INDEPENDENT;
 

base-commit: d998c62f267213aeb815cf654908608eb7c00db2

Signature · 1 line
2.53.0
#2 lore raw

[AMD Official Use Only - AMD Internal Distribution Only]

Quoted 25 lines

-----Original Message-----
From: Sean Christopherson <[email protected]>
Sent: Friday, April 3, 2026 2:36 PM
To: Kaplan, David <[email protected]>
Cc: Thomas Gleixner <[email protected]>; Ingo Molnar <[email protected]>; Borislav Petkov <[email protected]>; Dave Hansen <[email protected]>; [email protected]; H. Peter Anvin <[email protected]>; [email protected]
Subject: Re: [PATCH] x86/fpu: Disable shstk if no CET_USER state

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.

Quoted 10 lines · David Kaplan

Some hypervisors (including QEMU 10.1.5) may report CET_SS support in CPUID Fn7 but fail to report that CET_USER state is supported in supervisor xstate. Linux relies on XSAVES/XRSTORS to swap CET state during context switch and assumes it is supported when CET_SS is present.

As a result, if a user process is run with shadow stacks enabled and then is switched away from, the system may crash because the new process may be incorrectly run with shadow stacks enabled.

Quoted 7 lines
Quoted 2 lines

Detect this broken configuration and disable user shadow stacks unless CET_USER is supported in xstate.

It's not actually broken though, is it? Just "odd". AFAICT, neither the SDM nor the APM _requires_ CET_{U,S} to be supported in XSS if shadow stacks are
suppported.

The APM language at least isn't super clear on whether this is legal, but there are no real CPUs to my knowledge that do not enumerate XSS support for these. But this configuration is definitely broken for Linux, because it will eventually crash with random segfaults (and unfortunately this often takes minutes to manifest since very few processes actually try to run with shadow stacks apparently).

Quoted 23 lines

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int

Quoted 11 lines
---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
arch/x86/kernel/fpu/xstate.c +0 -0
index 76153dfb58c9..188323442b4d 100644
@@ -855,6 +855,17 @@ void __init fpu__init_system_xstate(unsigned int

legacy_size)

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size)

Quoted 11 lines
arch/x86/kernel/fpu/xstate.c +8 -0
              goto out_disable;
      }
 
+     if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+         !(fpu_kernel_cfg.max_features & XFEATURE_MASK_CET_USER)) {
+             /*
+              * The kernel relies on XSAVES/XRSTORS to context switch shadow
+              * stack state.  If this isn't present, disable user shadow
+              * stacks.
+              */
+             pr_err("x86/fpu: CET_USER not supported in xstate when CET is
Quoted 5 lines

supported. Disabling shadow stacks.\n");

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size) · line 866

Quoted 1 line
arch/x86/kernel/fpu/xstate.c +1 -0
+             setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);

Doesn't this apply to IBT as well? This code is also misplaced, as it needs to live after at least this code:

Good point, it likely does. I can't confirm that as I don't have IBT hardware, but assuming that a guest can see CET_IBT=1 this same problem would exist.

Quoted 11 lines

if (!cpu_feature_enabled(X86_FEATURE_XSAVES))
fpu_kernel_cfg.max_features &=
XFEATURE_MASK_USER_SUPPORTED;
else
fpu_kernel_cfg.max_features &=
XFEATURE_MASK_USER_SUPPORTED |

                                        XFEATURE_MASK_SUPERVISOR_SUPPORTED;

and should probably play nice with the "out_disable" path too.

Ok, that makes sense. I'm not sure if the out_disable path ever gets used, but wouldn't that cause other problems if XSAVE is fully disabled? I don't see it clearing any feature bits related to other XSTATE components.

--David Kaplan

Quoted 2 lines

All in all, setup_cet() seems like a much better fit, but unfortunately that runs before fpu__init_system() :-(

#3 lore raw

[AMD Official Use Only - AMD Internal Distribution Only]

Quoted 33 lines

-----Original Message-----
From: Kaplan, David
Sent: Friday, April 3, 2026 2:53 PM
To: 'Sean Christopherson' <[email protected]>
Cc: Thomas Gleixner <[email protected]>; Ingo Molnar <[email protected]>; Borislav Petkov <[email protected]>; Dave Hansen <[email protected]>; [email protected]; H. Peter Anvin <[email protected]>; [email protected]
Subject: RE: [PATCH] x86/fpu: Disable shstk if no CET_USER state

Quoted 21 lines

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int

Quoted 9 lines
---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
arch/x86/kernel/fpu/xstate.c +0 -0
index 76153dfb58c9..188323442b4d 100644
@@ -855,6 +855,17 @@ void __init fpu__init_system_xstate(unsigned int

legacy_size)

Quoted 11 lines
arch/x86/kernel/fpu/xstate.c +8 -0
              goto out_disable;
      }
 
+     if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+         !(fpu_kernel_cfg.max_features & XFEATURE_MASK_CET_USER)) {
+             /*
+              * The kernel relies on XSAVES/XRSTORS to context switch shadow
+              * stack state.  If this isn't present, disable user shadow
+              * stacks.
+              */
+             pr_err("x86/fpu: CET_USER not supported in xstate when CET is
Quoted 10 lines
Quoted 5 lines

supported. Disabling shadow stacks.\n");

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size) · line 866

Quoted 1 line
arch/x86/kernel/fpu/xstate.c +1 -0
+             setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);

Doesn't this apply to IBT as well? This code is also misplaced, as it needs to live after at least this code:

Good point, it likely does. I can't confirm that as I don't have IBT hardware, but assuming that a guest can see CET_IBT=1 this same problem would exist.

Actually, I don't think this does apply to IBT as well. Per Documentation/arch/x86/shstk.rst, only kernel IBT is currently supported by Linux. And kernel IBT does not require either CET_USER or CET_KERNEL XSS support from what I see. (CET_KERNEL is only for the shadow stack related MSRs)

--David Kaplan

#4 lore raw
Quoted 28 lines · David Kaplan
Quoted 28 lines

From: Kaplan, David

Quoted 26 lines

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int

Quoted 9 lines
---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
arch/x86/kernel/fpu/xstate.c +0 -0
index 76153dfb58c9..188323442b4d 100644
@@ -855,6 +855,17 @@ void __init fpu__init_system_xstate(unsigned int

legacy_size)

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size)

Quoted 11 lines
arch/x86/kernel/fpu/xstate.c +8 -0
              goto out_disable;
      }
 
+     if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+         !(fpu_kernel_cfg.max_features & XFEATURE_MASK_CET_USER)) {
+             /*
+              * The kernel relies on XSAVES/XRSTORS to context switch shadow
+              * stack state.  If this isn't present, disable user shadow
+              * stacks.
+              */
+             pr_err("x86/fpu: CET_USER not supported in xstate when CET is

supported. Disabling shadow stacks.\n");

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size) · line 866

Quoted 1 line
arch/x86/kernel/fpu/xstate.c +1 -0
+             setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);

Doesn't this apply to IBT as well? This code is also misplaced, as it needs to live after at least this code:

Quoted 8 lines
Quoted 2 lines

Good point, it likely does. I can't confirm that as I don't have IBT hardware, but assuming that a guest can see CET_IBT=1 this same problem would exist.

Actually, I don't think this does apply to IBT as well. Per Documentation/arch/x86/shstk.rst, only kernel IBT is currently supported by Linux. And kernel IBT does not require either CET_USER or CET_KERNEL XSS support from what I see. (CET_KERNEL is only for the shadow stack related MSRs)

KVM virtualizes IBT and SHSTK, for both user and kernel, and relies on the host kernel to save/restore IA32_U_CET.

Note, I think xsave_cpuid_features[] is also flawed. Per the SDM, {U,S}_CET also exist if IBT is supported:

Bit 20: CET_IBT. Supports CET indirect branch tracking features if 1. Processors that set this bit define bits 5:2 and bits 63:10 of the IA32_U_CET and IA32_S_CET MSRs.

The current code likely works because all "real" CPUs that support IBT also support SHSTK.

#5 lore raw

[AMD Official Use Only - AMD Internal Distribution Only]

Quoted 51 lines

-----Original Message-----
From: Sean Christopherson <[email protected]>
Sent: Monday, April 6, 2026 9:27 AM
To: Kaplan, David <[email protected]>
Cc: Thomas Gleixner <[email protected]>; Ingo Molnar <[email protected]>; Borislav Petkov <[email protected]>; Dave Hansen <[email protected]>; [email protected]; H. Peter Anvin <[email protected]>; [email protected]
Subject: Re: [PATCH] x86/fpu: Disable shstk if no CET_USER state

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.

Quoted 6 lines · David Kaplan
Quoted 6 lines

From: Kaplan, David

Quoted 5 lines

Review

Quoted 5 lines
---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
(quoted diff) +0 -0

b/arch/x86/kernel/fpu/xstate.c

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned

Quoted 4 lines

index 76153dfb58c9..188323442b4d 100644

arch/x86/kernel/fpu/xstate.c +0 -0
@@ -855,6 +855,17 @@ void __init fpu__init_system_xstate(unsigned

int

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

legacy_size)

Quoted 7 lines
arch/x86/kernel/fpu/xstate.c +4 -0
              goto out_disable;
      }
 
+     if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+         !(fpu_kernel_cfg.max_features & XFEATURE_MASK_CET_USER)) {
+             /*
+              * The kernel relies on XSAVES/XRSTORS to context switch

shadow

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size)

Quoted 4 lines
arch/x86/kernel/fpu/xstate.c +4 -0
+              * stack state.  If this isn't present, disable user shadow
+              * stacks.
+              */
+             pr_err("x86/fpu: CET_USER not supported in xstate when CET is

supported. Disabling shadow stacks.\n");

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size) · line 866

Quoted 1 line
arch/x86/kernel/fpu/xstate.c +1 -0
+             setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);

Doesn't this apply to IBT as well? This code is also misplaced, as it needs

to

Quoted 4 lines
Quoted 4 lines
Quoted 1 line

live after at least this code:

Good point, it likely does. I can't confirm that as I don't have IBT hardware, but assuming that a guest can see CET_IBT=1 this same problem would

exist.

Quoted 1 line
Quoted 8 lines
Quoted 5 lines

Actually, I don't think this does apply to IBT as well. Per Documentation/arch/x86/shstk.rst, only kernel IBT is currently supported by Linux. And kernel IBT does not require either CET_USER or CET_KERNEL XSS support from what I see. (CET_KERNEL is only for the shadow stack related MSRs)

KVM virtualizes IBT and SHSTK, for both user and kernel, and relies on the host kernel to save/restore IA32_U_CET.

I think you're talking about a nested virt scenario is that right? So KVM in L1 virtualizes IBT/SHSTK for L2 and relies on the L1 kernel to save/restore IA32_UCET? And the concern is that if L1 doesn't support the XSS components then this is broken.

But it looks like kvm_setup_xss_caps() appears to check if host XSS includes all the CET components and will clear X86_FEATURE_SHSTK/IBT if they aren't present. So I think what you'd see is:

L0: CPUID says SHSTK/IBT supported. XSS bits supported.
L1: CPUID says SHSTK/IBT supported. XSS bits not supported
L2: CPUID doesn't say SHSTK/IBT supported.

Or did I misunderstand the scenario you're talking about?

Quoted 11 lines

Note, I think xsave_cpuid_features[] is also flawed. Per the SDM, {U,S}_CET also
exist if IBT is supported:

Bit 20: CET_IBT. Supports CET indirect branch tracking features if 1.
Processors
that set this bit define bits 5:2 and bits 63:10 of the IA32_U_CET and
IA32_S_CET
MSRs.

Quoted 3 lines

The current code likely works because all "real" CPUs that support IBT also support
SHSTK.

Yeah that also looks incorrect, but benign (at least for now).

--David Kaplan

#6 lore raw
Quoted 44 lines · David Kaplan
Quoted 44 lines

From: Sean Christopherson <[email protected]>

Quoted 6 lines · David Kaplan
Quoted 6 lines

From: Kaplan, David

Quoted 5 lines

Review

Quoted 5 lines
---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
(quoted diff) +0 -0

b/arch/x86/kernel/fpu/xstate.c

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned

Quoted 4 lines

index 76153dfb58c9..188323442b4d 100644

arch/x86/kernel/fpu/xstate.c +0 -0
@@ -855,6 +855,17 @@ void __init fpu__init_system_xstate(unsigned

int

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

legacy_size)

Quoted 7 lines
arch/x86/kernel/fpu/xstate.c +4 -0
              goto out_disable;
      }
 
+     if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+         !(fpu_kernel_cfg.max_features & XFEATURE_MASK_CET_USER)) {
+             /*
+              * The kernel relies on XSAVES/XRSTORS to context switch

shadow

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size)

Quoted 4 lines
arch/x86/kernel/fpu/xstate.c +4 -0
+              * stack state.  If this isn't present, disable user shadow
+              * stacks.
+              */
+             pr_err("x86/fpu: CET_USER not supported in xstate when CET is

supported. Disabling shadow stacks.\n");

arch/x86/kernel/fpu/xstate.c void __init fpu__init_system_xstate(unsigned int legacy_size) · line 866

Quoted 1 line
arch/x86/kernel/fpu/xstate.c +1 -0
+             setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);

Doesn't this apply to IBT as well? This code is also misplaced, as it needs

to

Quoted 4 lines
Quoted 4 lines
Quoted 1 line

live after at least this code:

Good point, it likely does. I can't confirm that as I don't have IBT hardware, but assuming that a guest can see CET_IBT=1 this same problem would

exist.

Quoted 6 lines

Actually, I don't think this does apply to IBT as well. Per Documentation/arch/x86/shstk.rst, only kernel IBT is currently supported by Linux. And kernel IBT does not require either CET_USER or CET_KERNEL XSS support from what I see. (CET_KERNEL is only for the shadow stack related MSRs)

Quoted 4 lines
Quoted 2 lines

KVM virtualizes IBT and SHSTK, for both user and kernel, and relies on the host kernel to save/restore IA32_U_CET.

I think you're talking about a nested virt scenario is that right?

FWIW, this isn't limited to running in a VM. Booting on bare metal with e.g. noxsaves=1 would lead to the same problematic scenario.

Quoted 5 lines

So KVM in L1 virtualizes IBT/SHSTK for L2 and relies on the L1 kernel to save/restore IA32_UCET? And the concern is that if L1 doesn't support the XSS components then this is broken.

But it looks like kvm_setup_xss_caps() appears to check if host XSS includes

Gah, -ENOCOFFEE, I forgot that KVM manually verifies that the kernel will context switch CET.

Quoted 8 lines

all the CET components and will clear X86_FEATURE_SHSTK/IBT if they aren't present. So I think what you'd see is:

L0: CPUID says SHSTK/IBT supported. XSS bits supported.
L1: CPUID says SHSTK/IBT supported. XSS bits not supported
L2: CPUID doesn't say SHSTK/IBT supported.

Or did I misunderstand the scenario you're talking about?

#7 lore raw

[AMD Official Use Only - AMD Internal Distribution Only]

Quoted 67 lines

-----Original Message-----
From: Sean Christopherson <[email protected]>
Sent: Monday, April 6, 2026 10:32 AM
To: Kaplan, David <[email protected]>
Cc: Thomas Gleixner <[email protected]>; Ingo Molnar <[email protected]>; Borislav Petkov <[email protected]>; Dave Hansen <[email protected]>; [email protected]; H. Peter Anvin <[email protected]>; [email protected]
Subject: Re: [PATCH] x86/fpu: Disable shstk if no CET_USER state

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.

Quoted 7 lines · David Kaplan
Quoted 7 lines

From: Sean Christopherson <[email protected]>

Quoted 5 lines · David Kaplan
Quoted 5 lines

From: Kaplan, David

Quoted 4 lines
Quoted 4 lines
---
 arch/x86/kernel/fpu/xstate.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
Quoted 6 lines
Quoted 6 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line

Review

Quoted 1 line
(quoted diff) +0 -0

b/arch/x86/kernel/fpu/xstate.c

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

arch/x86/kernel/fpu/xstate.c void __init

Quoted 4 lines

index 76153dfb58c9..188323442b4d 100644

arch/x86/kernel/fpu/xstate.c +0 -0
@@ -855,6 +855,17 @@ void __init

fpu__init_system_xstate(unsigned

Quoted 7 lines
Quoted 7 lines

int

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

legacy_size)

Quoted 5 lines
arch/x86/kernel/fpu/xstate.c +2 -0
              goto out_disable;
      }
 
+     if (boot_cpu_has(X86_FEATURE_USER_SHSTK) &&
+         !(fpu_kernel_cfg.max_features &

XFEATURE_MASK_CET_USER)) {

Quoted 7 lines
Quoted 7 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

arch/x86/kernel/fpu/xstate.c void __init · line 861

Quoted 2 lines
arch/x86/kernel/fpu/xstate.c +2 -0
+             /*
+              * The kernel relies on XSAVES/XRSTORS to context switch

shadow

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
arch/x86/kernel/fpu/xstate.c +4 -0
+              * stack state.  If this isn't present, disable user shadow
+              * stacks.
+              */
+             pr_err("x86/fpu: CET_USER not supported in xstate when

CET is

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

supported. Disabling shadow stacks.\n");

arch/x86/kernel/fpu/xstate.c void __init · line 866

Quoted 1 line
arch/x86/kernel/fpu/xstate.c +1 -0
+             setup_clear_cpu_cap(X86_FEATURE_USER_SHSTK);

Doesn't this apply to IBT as well? This code is also misplaced, as it

needs

Quoted 4 lines
Quoted 4 lines

to

Quoted 3 lines
Quoted 3 lines
Quoted 1 line

live after at least this code:

Good point, it likely does. I can't confirm that as I don't have IBT

hardware,

Quoted 5 lines
Quoted 5 lines
Quoted 1 line
Quoted 1 line

but assuming that a guest can see CET_IBT=1 this same problem would

exist.

Quoted 3 lines

Actually, I don't think this does apply to IBT as well. Per Documentation/arch/x86/shstk.rst, only kernel IBT is currently

supported by

Quoted 1 line
Quoted 1 line
Quoted 1 line

Linux. And kernel IBT does not require either CET_USER or CET_KERNEL

XSS

Quoted 1 line
Quoted 1 line
Quoted 1 line

support from what I see. (CET_KERNEL is only for the shadow stack

related

Quoted 2 lines
Quoted 2 lines
Quoted 1 line

MSRs)

Quoted 9 lines
Quoted 1 line
Quoted 1 line

KVM virtualizes IBT and SHSTK, for both user and kernel, and relies on the

host

Quoted 3 lines
Quoted 1 line

kernel to save/restore IA32_U_CET.

I think you're talking about a nested virt scenario is that right?

FWIW, this isn't limited to running in a VM. Booting on bare metal with e.g. noxsaves=1 would lead to the same problematic scenario.

Btw, this does not appear to be a problem. X86_FEATURE_SHSTK is marked as dependent on X86_FEATURE_XSAVES so booting with noxsaves seems to work fine.

--David Kaplan