Skip to contentSkip to conversation

x86: clear XD_DISABLED flag on Intel to regain NX

PATCH Superseded Kees Cook 3 participants 4 messages Jun 17, 2010 – Jun 18, 2010

Subsystems (2)

v1 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Kees Cook

author
#0 lore raw

PATCH 1/1 x86: clear XD_DISABLED flag on Intel to regain NX

This will clear the MSR_IA32_MISC_ENABLE_XD_DISABLE bit so that NX cannot be inappropriately controlled by the BIOS on Intel CPUs. If NX actually needs to be disabled, "noexec=off" can be used.

---
 arch/x86/kernel/head_32.S |   19 +++++++++++++++++++
 arch/x86/kernel/head_64.S |   18 ++++++++++++++++++
 arch/x86/mm/setup_nx.c    |    2 +-
 3 files changed, 38 insertions(+), 1 deletions(-)
arch/x86/kernel/head_32.S +19 -0
index 37c3d4b..111e434 100644
@@ -309,6 +309,25 @@ ENTRY(startup_32_smp)
 	subl $0x80000001, %eax
 	cmpl $(0x8000ffff-0x80000001), %eax
 	ja 6f
+
+	/* Is this "GenuineIntel"? */
+	movl $0x0, %eax
+	cpuid
+	cmpl $0x756e6547, %ebx
+	jnz 5f
+	cmpl $0x49656e69, %edx
+	jnz 5f
+	cmpl $0x6c65746e, %ecx
+	jnz 5f
+
+	/* Clear MSR_IA32_MISC_ENABLE_XD_DISABLE if set */
+	movl $MSR_IA32_MISC_ENABLE, %ecx
+	rdmsr
+	btrl $2, %edx
+	jnc 5f
+	wrmsr
+
+5:
 	mov $0x80000001, %eax
 	cpuid
 	/* Execute Disable bit supported? */
arch/x86/kernel/head_64.S +18 -0
index 3d1e6f1..e918e40 100644
@@ -175,6 +175,24 @@ ENTRY(secondary_startup_64)
 	jmp	*%rax
 1:
 
+	/* Is this "GenuineIntel"? */
+	movl $0x0, %eax
+	cpuid
+	cmpl $0x756e6547, %ebx
+	jnz 5f
+	cmpl $0x49656e69, %edx
+	jnz 5f
+	cmpl $0x6c65746e, %ecx
+	jnz 5f
+
+	/* Clear MSR_IA32_MISC_ENABLE_XD_DISABLE if set */
+	movl $MSR_IA32_MISC_ENABLE, %ecx
+	rdmsr
+	btrl $2, %edx
+	jnc 5f
+	wrmsr
+
+5:
 	/* Check if nx is implemented */
 	movl	$0x80000001, %eax
 	cpuid
arch/x86/mm/setup_nx.c +1 -1
index a3250aa..410531d 100644
@@ -41,7 +41,7 @@ void __init x86_report_nx(void)
 {
 	if (!cpu_has_nx) {
 		printk(KERN_NOTICE "Notice: NX (Execute Disable) protection "
-		       "missing in CPU or disabled in BIOS!\n");
+		       "missing in CPU!\n");
 	} else {
 #if defined(CONFIG_X86_64) || defined(CONFIG_X86_PAE)
 		if (disable_nx) {
Signature · 5 lines
1.7.1

-- 
Kees Cook
Ubuntu Security Team
#1 lore raw
Quoted 31 lines · Kees Cook

This will clear the MSR_IA32_MISC_ENABLE_XD_DISABLE bit so that NX cannot be inappropriately controlled by the BIOS on Intel CPUs. If NX actually needs to be disabled, "noexec=off" can be used.

---
 arch/x86/kernel/head_32.S |   19 +++++++++++++++++++
 arch/x86/kernel/head_64.S |   18 ++++++++++++++++++
 arch/x86/mm/setup_nx.c    |    2 +-
 3 files changed, 38 insertions(+), 1 deletions(-)
arch/x86/kernel/head_32.S +12 -0
index 37c3d4b..111e434 100644
@@ -309,6 +309,25 @@ ENTRY(startup_32_smp)
 	subl $0x80000001, %eax
 	cmpl $(0x8000ffff-0x80000001), %eax
 	ja 6f
+
+	/* Is this "GenuineIntel"? */
+	movl $0x0, %eax
+	cpuid
+	cmpl $0x756e6547, %ebx
+	jnz 5f
+	cmpl $0x49656e69, %edx
+	jnz 5f
+	cmpl $0x6c65746e, %ecx
+	jnz 5f
+
+	/* Clear MSR_IA32_MISC_ENABLE_XD_DISABLE if set */

arch/x86/kernel/head_32.S ENTRY(startup_32_smp) · line 333

Quoted 10 lines
arch/x86/kernel/head_32.S +7 -0
+	movl $MSR_IA32_MISC_ENABLE, %ecx
+	rdmsr
+	btrl $2, %edx
+	jnc 5f
+	wrmsr
+
+5:
 	mov $0x80000001, %eax
 	cpuid
 	/* Execute Disable bit supported? */

Multiple problems with this code.

a) Not all Intel CPUs with extended CPUID levels have MSR_IA32_MISC_ENABLE bit 34. Since we can't take traps here we would have to know positively that we aren't going to trip on anything.
b) For 64 bits, this should go into verify_cpu_64.S, and since that is 32-bit code anyway, it would be best if we could merge the 32- and 64-bit code into that file; it already simply returns a value that could be ignored on 32 bits.

-hpa

#2 lore raw
Quoted 4 lines

a) Not all Intel CPUs with extended CPUID levels have MSR_IA32_MISC_ENABLE bit 34. Since we can't take traps here we would have to know positively that we aren't going to trip on anything.

It won't as these older Intel CPUs are always going to read this bit as zero, and even if no checking for this is include all the wrmsr would do is write the same value back, which won't cause a trap.

Yuhong Bao

_________________________________________________________________
Hotmail has tools for the New Busy. Search, chat and e-mail from your inbox.
http://www.windowslive.com/campaign/thenewbusy?ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_1

#3 lore raw

Yes, I mentally flipped the sense of the bit... talked to Kees already.

Quoted 11 lines · Yuhong Bao
Quoted 4 lines

a) Not all Intel CPUs with extended CPUID levels have MSR_IA32_MISC_ENABLE bit 34. Since we can't take traps here we would have to know positively that we aren't going to trip on anything.

It won't as these older Intel CPUs are always going to read this bit as zero, and even if no checking for this is include all the wrmsr would do is write the same value back, which won't cause a trap.

Yuhong Bao

_________________________________________________________________
Hotmail has tools for the New Busy. Search, chat and e-mail from your inbox.
http://www.windowslive.com/campaign/thenewbusy?ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_1

Signature · 1 line
Sent from my mobile phone.  Please pardon any lack of formatting.