]> git.proxmox.com Git - pve-kernel.git/commitdiff
KPTI: disable on AMD
authorFabian Grünbichler <f.gruenbichler@proxmox.com>
Mon, 8 Jan 2018 09:25:31 +0000 (10:25 +0100)
committerFabian Grünbichler <f.gruenbichler@proxmox.com>
Mon, 8 Jan 2018 09:25:31 +0000 (10:25 +0100)
and allow loading of microcode on recent AMD systems in preparation of
further Spectre fixes

patches/kernel/0240-x86-cpu-x86-pti-Do-not-enable-PTI-on-AMD-processors.patch [new file with mode: 0644]
patches/kernel/0241-x86-microcode-AMD-Add-support-for-fam17h-microcode-l.patch [new file with mode: 0644]

diff --git a/patches/kernel/0240-x86-cpu-x86-pti-Do-not-enable-PTI-on-AMD-processors.patch b/patches/kernel/0240-x86-cpu-x86-pti-Do-not-enable-PTI-on-AMD-processors.patch
new file mode 100644 (file)
index 0000000..6499a52
--- /dev/null
@@ -0,0 +1,54 @@
+From 5462db3d070845ecc34929b6f25a87efda023aae Mon Sep 17 00:00:00 2001
+From: Tom Lendacky <thomas.lendacky@amd.com>
+Date: Tue, 26 Dec 2017 23:43:54 -0600
+Subject: [PATCH 240/241] x86/cpu, x86/pti: Do not enable PTI on AMD processors
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+CVE-2017-5754
+
+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
+
+(cherry picked from commit 694d99d40972f12e59a3696effee8a376b79d7c8)
+Signed-off-by: Marcelo Henrique Cerri <marcelo.cerri@canonical.com>
+(cherry picked from commit 9d334f48f017b9c6457c6ba321e5a53a1cc6a5c7)
+Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.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 99f37d1636ff..1854dd8071a6 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);
+ }
+-- 
+2.14.2
+
diff --git a/patches/kernel/0241-x86-microcode-AMD-Add-support-for-fam17h-microcode-l.patch b/patches/kernel/0241-x86-microcode-AMD-Add-support-for-fam17h-microcode-l.patch
new file mode 100644 (file)
index 0000000..6986dec
--- /dev/null
@@ -0,0 +1,52 @@
+From 8329d47141a78a64e8ae6f4a735aceaafe93e098 Mon Sep 17 00:00:00 2001
+From: Tom Lendacky <thomas.lendacky@amd.com>
+Date: Thu, 30 Nov 2017 16:46:40 -0600
+Subject: [PATCH 241/241] x86/microcode/AMD: Add support for fam17h microcode
+ loading
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+commit f4e9b7af0cd58dd039a0fb2cd67d57cea4889abf upstream.
+
+The size for the Microcode Patch Block (MPB) for an AMD family 17h
+processor is 3200 bytes.  Add a #define for fam17h so that it does
+not default to 2048 bytes and fail a microcode load/update.
+
+Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
+Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
+Reviewed-by: Borislav Petkov <bp@alien8.de>
+Link: https://lkml.kernel.org/r/20171130224640.15391.40247.stgit@tlendack-t1.amdoffice.net
+Signed-off-by: Ingo Molnar <mingo@kernel.org>
+Cc: Alice Ferrazzi <alicef@gentoo.org>
+Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
+Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
+---
+ arch/x86/kernel/cpu/microcode/amd.c | 4 ++++
+ 1 file changed, 4 insertions(+)
+
+diff --git a/arch/x86/kernel/cpu/microcode/amd.c b/arch/x86/kernel/cpu/microcode/amd.c
+index 21b185793c80..248cad00fee6 100644
+--- a/arch/x86/kernel/cpu/microcode/amd.c
++++ b/arch/x86/kernel/cpu/microcode/amd.c
+@@ -467,6 +467,7 @@ static unsigned int verify_patch_size(u8 family, u32 patch_size,
+ #define F14H_MPB_MAX_SIZE 1824
+ #define F15H_MPB_MAX_SIZE 4096
+ #define F16H_MPB_MAX_SIZE 3458
++#define F17H_MPB_MAX_SIZE 3200
+       switch (family) {
+       case 0x14:
+@@ -478,6 +479,9 @@ static unsigned int verify_patch_size(u8 family, u32 patch_size,
+       case 0x16:
+               max_size = F16H_MPB_MAX_SIZE;
+               break;
++      case 0x17:
++              max_size = F17H_MPB_MAX_SIZE;
++              break;
+       default:
+               max_size = F1XH_MPB_MAX_SIZE;
+               break;
+-- 
+2.14.2
+