Skip to contentSkip to conversation

x86/spectre: Expand test for vulnerability to empty RSB exploits

PATCH In review Jim Mattson 3 participants 5 messages Aug 7, 2018 – Aug 20, 2018

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/spectre: Expand test for vulnerability to empty RSB exploits

Skylake-era Intel CPUs are vulnerable to exploits of empty RSB conditions. On hardware, platform vulnerability can be determined simply by checking the processor's DisplayModel/DisplayFamily signature. However, when running in a VM, the operating system should also query IA32_ARCH_CAPABILITIES.RSBA[bit 2], a synthetic bit that can be set by a hypervisor to indicate that the VM might run on a vulnerable physical processor, regardless of the DisplayModel/DisplayFamily reported by CPUID.

Note that IA32_ARCH_CAPABILITIES.RSBA[bit 2] is always clear on hardware, so the DisplayModel/DisplayFamily check is still required.

For all of the details, see the Intel white paper, "Retpoline: A Branch Target Injection Mitigation" (document number 337131-001), section 5.3: Virtual Machine CPU Identification.

---
 arch/x86/include/asm/msr-index.h |  1 +
 arch/x86/kernel/cpu/bugs.c       | 14 +++++++++++++-
 2 files changed, 14 insertions(+), 1 deletion(-)
arch/x86/include/asm/msr-index.h +1 -0
index 68b2c3150de1..f37ec58c4e04 100644
@@ -70,6 +70,7 @@
 #define MSR_IA32_ARCH_CAPABILITIES	0x0000010a
 #define ARCH_CAP_RDCL_NO		(1 << 0)   /* Not susceptible to Meltdown */
 #define ARCH_CAP_IBRS_ALL		(1 << 1)   /* Enhanced IBRS support */
+#define ARCH_CAP_RSBA			(1 << 2)   /* Vulnerable to empty RSB */
 #define ARCH_CAP_SSB_NO			(1 << 4)   /*
 						    * Not susceptible to Speculative Store Bypass
 						    * attack, so no Speculative Store Bypass
arch/x86/kernel/cpu/bugs.c +13 -1
index 5c0ea39311fe..b6fe335746a4 100644
@@ -330,6 +330,18 @@ static bool __init is_skylake_era(void)
 	return false;
 }
 
+/* Check for vulnerability to exploits of empty RSB conditions */
+static bool __init is_vulnerable_to_empty_rsb(void)
+{
+	u64 ia32_cap = 0;
+
+	if (boot_cpu_has(X86_FEATURE_ARCH_CAPABILITIES))
+		rdmsrl(MSR_IA32_ARCH_CAPABILITIES, ia32_cap);
+
+	return (ia32_cap & ARCH_CAP_RSBA) || is_skylake_era();
+}
+
+
 static void __init spectre_v2_select_mitigation(void)
 {
 	enum spectre_v2_mitigation_cmd cmd = spectre_v2_parse_cmdline();
@@ -402,7 +414,7 @@ static void __init spectre_v2_select_mitigation(void)
 	 * switch is required.
 	 */
 	if ((!boot_cpu_has(X86_FEATURE_PTI) &&
-	     !boot_cpu_has(X86_FEATURE_SMEP)) || is_skylake_era()) {
+	     !boot_cpu_has(X86_FEATURE_SMEP)) || is_vulnerable_to_empty_rsb()) {
 		setup_force_cpu_cap(X86_FEATURE_RSB_CTXSW);
 		pr_info("Spectre v2 mitigation: Filling RSB on context switch\n");
 	}
Signature · 1 line
2.18.0.597.ga71716f1ad-goog
Reviewed-by on v1 1/1
#1 lore raw
Quoted 12 lines · Jim Mattson

Skylake-era Intel CPUs are vulnerable to exploits of empty RSB conditions. On hardware, platform vulnerability can be determined simply by checking the processor's DisplayModel/DisplayFamily signature. However, when running in a VM, the operating system should also query IA32_ARCH_CAPABILITIES.RSBA[bit 2], a synthetic bit that can be set by a hypervisor to indicate that the VM might run on a vulnerable physical processor, regardless of the DisplayModel/DisplayFamily reported by CPUID.

Note that IA32_ARCH_CAPABILITIES.RSBA[bit 2] is always clear on hardware, so the DisplayModel/DisplayFamily check is still required.

Quoted 6 lines

For all of the details, see the Intel white paper, "Retpoline: A Branch Target Injection Mitigation" (document number 337131-001), section 5.3: Virtual Machine CPU Identification.

Thank you as it saves me from doing this :-)

Quoted 52 lines
---
 arch/x86/include/asm/msr-index.h |  1 +
 arch/x86/kernel/cpu/bugs.c       | 14 +++++++++++++-
 2 files changed, 14 insertions(+), 1 deletion(-)
arch/x86/include/asm/msr-index.h +1 -0
index 68b2c3150de1..f37ec58c4e04 100644
@@ -70,6 +70,7 @@
 #define MSR_IA32_ARCH_CAPABILITIES	0x0000010a
 #define ARCH_CAP_RDCL_NO		(1 << 0)   /* Not susceptible to Meltdown */
 #define ARCH_CAP_IBRS_ALL		(1 << 1)   /* Enhanced IBRS support */
+#define ARCH_CAP_RSBA			(1 << 2)   /* Vulnerable to empty RSB */
 #define ARCH_CAP_SSB_NO			(1 << 4)   /*
 						    * Not susceptible to Speculative Store Bypass
 						    * attack, so no Speculative Store Bypass
arch/x86/kernel/cpu/bugs.c +13 -1
index 5c0ea39311fe..b6fe335746a4 100644
@@ -330,6 +330,18 @@ static bool __init is_skylake_era(void)
 	return false;
 }
 
+/* Check for vulnerability to exploits of empty RSB conditions */
+static bool __init is_vulnerable_to_empty_rsb(void)
+{
+	u64 ia32_cap = 0;
+
+	if (boot_cpu_has(X86_FEATURE_ARCH_CAPABILITIES))
+		rdmsrl(MSR_IA32_ARCH_CAPABILITIES, ia32_cap);
+
+	return (ia32_cap & ARCH_CAP_RSBA) || is_skylake_era();
+}
+
+
 static void __init spectre_v2_select_mitigation(void)
 {
 	enum spectre_v2_mitigation_cmd cmd = spectre_v2_parse_cmdline();
@@ -402,7 +414,7 @@ static void __init spectre_v2_select_mitigation(void)
 	 * switch is required.
 	 */
 	if ((!boot_cpu_has(X86_FEATURE_PTI) &&
-	     !boot_cpu_has(X86_FEATURE_SMEP)) || is_skylake_era()) {
+	     !boot_cpu_has(X86_FEATURE_SMEP)) || is_vulnerable_to_empty_rsb()) {
 		setup_force_cpu_cap(X86_FEATURE_RSB_CTXSW);
 		pr_info("Spectre v2 mitigation: Filling RSB on context switch\n");
 	}
Signature · 1 line
2.18.0.597.ga71716f1ad-goog
#2 lore raw
In reply to Jim Mattson
Quoted 12 lines · Jim Mattson

Skylake-era Intel CPUs are vulnerable to exploits of empty RSB conditions. On hardware, platform vulnerability can be determined simply by checking the processor's DisplayModel/DisplayFamily signature. However, when running in a VM, the operating system should also query IA32_ARCH_CAPABILITIES.RSBA[bit 2], a synthetic bit that can be set by a hypervisor to indicate that the VM might run on a vulnerable physical processor, regardless of the DisplayModel/DisplayFamily reported by CPUID.

Note that IA32_ARCH_CAPABILITIES.RSBA[bit 2] is always clear on hardware, so the DisplayModel/DisplayFamily check is still required.

Quoted 6 lines

For all of the details, see the Intel white paper, "Retpoline: A Branch Target Injection Mitigation" (document number 337131-001), section 5.3: Virtual Machine CPU Identification.

That has been superseeded by:

fdf82a7856b3 ("x86/speculation: Protect against userspace-userspace spectreRSB")

right? At least it does not apply anymore...

Thanks,

tglx

#3 lore raw
Quoted 19 lines · Thomas Gleixner
Quoted 16 lines · Jim Mattson

Skylake-era Intel CPUs are vulnerable to exploits of empty RSB conditions. On hardware, platform vulnerability can be determined simply by checking the processor's DisplayModel/DisplayFamily signature. However, when running in a VM, the operating system should also query IA32_ARCH_CAPABILITIES.RSBA[bit 2], a synthetic bit that can be set by a hypervisor to indicate that the VM might run on a vulnerable physical processor, regardless of the DisplayModel/DisplayFamily reported by CPUID.

Note that IA32_ARCH_CAPABILITIES.RSBA[bit 2] is always clear on hardware, so the DisplayModel/DisplayFamily check is still required.

For all of the details, see the Intel white paper, "Retpoline: A Branch Target Injection Mitigation" (document number 337131-001), section 5.3: Virtual Machine CPU Identification.

Quoted 8 lines
Quoted 2 lines

That has been superseeded by:

fdf82a7856b3 ("x86/speculation: Protect against userspace-userspace spectreRSB")

right? At least it does not apply anymore...

Right. It doesn't appear that Skylake CPUs get any special treatment any more.

#4 lore raw
Quoted 23 lines · Jim Mattson
Quoted 22 lines · Thomas Gleixner
Quoted 18 lines · Jim Mattson

Skylake-era Intel CPUs are vulnerable to exploits of empty RSB conditions. On hardware, platform vulnerability can be determined simply by checking the processor's DisplayModel/DisplayFamily signature. However, when running in a VM, the operating system should also query IA32_ARCH_CAPABILITIES.RSBA[bit 2], a synthetic bit that can be set by a hypervisor to indicate that the VM might run on a vulnerable physical processor, regardless of the DisplayModel/DisplayFamily reported by CPUID.

Note that IA32_ARCH_CAPABILITIES.RSBA[bit 2] is always clear on hardware, so the DisplayModel/DisplayFamily check is still required.

For all of the details, see the Intel white paper, "Retpoline: A Branch Target Injection Mitigation" (document number 337131-001), section 5.3: Virtual Machine CPU Identification.

Quoted 7 lines
Quoted 5 lines

That has been superseeded by:

fdf82a7856b3 ("x86/speculation: Protect against userspace-userspace spectreRSB")

right? At least it does not apply anymore...

Right. It doesn't appear that Skylake CPUs get any special treatment any more.

Yeah, it's universally f*cked up by now.

Thanks,

tglx