Merge branch 'x86-microcode-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 microcode update from Ingo Molnar:
 "The biggest change (by Borislav Petkov) is a thorough rewrite of the
  Intel microcode loader and its interactions with the core code.

  The biggest conceptual change is the decoupling of the microcode
  loading on boot and application processors (which load the microcode
  in different scenarios), so that both parse the input patches with as
  few assumptions as possible - this also fixes various kernel address
  space randomization bugs. (The AP side then goes on and caches the
  result to improve boot performance.)

  Since the AMD side already did this, this change also opened up the
  path towards more unification/simplification of the core microcode
  loading infrastructure:

     10 files changed, 647 insertions(+), 940 deletions(-)

  which speaks for itself"

* 'x86-microcode-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/microcode: Bump driver version, update copyrights
  x86/microcode: Rework microcode loading
  x86/microcode/intel: Remove intel_lib.c
  x86/microcode/amd: Move private inlines to .c and mark local functions static
  x86/microcode: Collect CPU info on resume
  x86/microcode: Issue the debug printk on resume only on success
  x86/microcode/amd: Hand down the CPU family
  x86/microcode: Export the microcode cache linked list
  x86/microcode: Remove one #ifdef clause
  x86/microcode/intel: Simplify generic_load_microcode()
  x86/microcode: Move driver authors to CREDITS
  x86/microcode: Run the AP-loading routine only on the application processors

10 files changed, 647 insertions, 940 deletions

diff --git a/CREDITS b/CREDITS
index d7ebdfbc4d4f..10a9eee807b6 100644
--- a/CREDITS
+++ b/CREDITS
@@ -2775,6 +2775,10 @@ S: C/ Mieses 20, 9-B
 S: Valladolid 47009
 S: Spain
 
+N: Peter Oruba
+D: AMD Microcode loader driver
+S: Germany
+
 N: Jens Osterkamp
 E: jens@de.ibm.com
 D: Maintainer of Spidernet network driver for Cell
diff --git a/arch/x86/include/asm/microcode.h b/arch/x86/include/asm/microcode.h
index da0d81fa0b54..38711df3bcb5 100644
--- a/arch/x86/include/asm/microcode.h
+++ b/arch/x86/include/asm/microcode.h
@@ -20,6 +20,15 @@ do {							\
 			 (u32)((u64)(val)),		\
 			 (u32)((u64)(val) >> 32))
 
+struct ucode_patch {
+	struct list_head plist;
+	void *data;		/* Intel uses only this one */
+	u32 patch_id;
+	u16 equiv_cpu;
+};
+
+extern struct list_head microcode_cache;
+
 struct cpu_signature {
 	unsigned int sig;
 	unsigned int pf;
@@ -55,12 +64,7 @@ struct ucode_cpu_info {
 	void			*mc;
 };
 extern struct ucode_cpu_info ucode_cpu_info[];
-
-#ifdef CONFIG_MICROCODE
-int __init microcode_init(void);
-#else
-static inline int __init microcode_init(void)	{ return 0; };
-#endif
+struct cpio_data find_microcode_in_initrd(const char *path, bool use_pa);
 
 #ifdef CONFIG_MICROCODE_INTEL
 extern struct microcode_ops * __init init_intel_microcode(void);
@@ -131,11 +135,13 @@ static inline unsigned int x86_cpuid_family(void)
 }
 
 #ifdef CONFIG_MICROCODE
+int __init microcode_init(void);
 extern void __init load_ucode_bsp(void);
 extern void load_ucode_ap(void);
 void reload_early_microcode(void);
 extern bool get_builtin_firmware(struct cpio_data *cd, const char *name);
 #else
+static inline int __init microcode_init(void)			{ return 0; };
 static inline void __init load_ucode_bsp(void)			{ }
 static inline void load_ucode_ap(void)				{ }
 static inline void reload_early_microcode(void)			{ }
diff --git a/arch/x86/include/asm/microcode_amd.h b/arch/x86/include/asm/microcode_amd.h
index 15eb75484cc0..3e3e20be829a 100644
--- a/arch/x86/include/asm/microcode_amd.h
+++ b/arch/x86/include/asm/microcode_amd.h
@@ -40,38 +40,18 @@ struct microcode_amd {
 	unsigned int			mpb[0];
 };
 
-static inline u16 find_equiv_id(struct equiv_cpu_entry *equiv_cpu_table,
-				unsigned int sig)
-{
-	int i = 0;
-
-	if (!equiv_cpu_table)
-		return 0;
-
-	while (equiv_cpu_table[i].installed_cpu != 0) {
-		if (sig == equiv_cpu_table[i].installed_cpu)
-			return equiv_cpu_table[i].equiv_cpu;
-
-		i++;
-	}
-	return 0;
-}
-
-extern int __apply_microcode_amd(struct microcode_amd *mc_amd);
-extern int apply_microcode_amd(int cpu);
-extern enum ucode_state load_microcode_amd(int cpu, u8 family, const u8 *data, size_t size);
-
 #define PATCH_MAX_SIZE PAGE_SIZE
 
 #ifdef CONFIG_MICROCODE_AMD
 extern void __init load_ucode_amd_bsp(unsigned int family);
-extern void load_ucode_amd_ap(void);
-extern int __init save_microcode_in_initrd_amd(void);
+extern void load_ucode_amd_ap(unsigned int family);
+extern int __init save_microcode_in_initrd_amd(unsigned int family);
 void reload_ucode_amd(void);
 #else
 static inline void __init load_ucode_amd_bsp(unsigned int family) {}
-static inline void load_ucode_amd_ap(void) {}
-static inline int __init save_microcode_in_initrd_amd(void) { return -EINVAL; }
+static inline void load_ucode_amd_ap(unsigned int family) {}
+static inline int __init
+save_microcode_in_initrd_amd(unsigned int family) { return -EINVAL; }
 void reload_ucode_amd(void) {}
 #endif
 
diff --git a/arch/x86/include/asm/microcode_intel.h b/arch/x86/include/asm/microcode_intel.h
index 5e69154c9f07..195becc6f780 100644
--- a/arch/x86/include/asm/microcode_intel.h
+++ b/arch/x86/include/asm/microcode_intel.h
@@ -52,10 +52,6 @@ struct extended_sigtable {
 
 #define exttable_size(et) ((et)->count * EXT_SIGNATURE_SIZE + EXT_HEADER_SIZE)
 
-extern int has_newer_microcode(void *mc, unsigned int csig, int cpf, int rev);
-extern int microcode_sanity_check(void *mc, int print_err);
-extern int find_matching_signature(void *mc, unsigned int csig, int cpf);
-
 #ifdef CONFIG_MICROCODE_INTEL
 extern void __init load_ucode_intel_bsp(void);
 extern void load_ucode_intel_ap(void);
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 87f563946e6b..729f92ba8224 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -1436,11 +1436,8 @@ void cpu_init(void)
 	 */
 	cr4_init_shadow();
 
-	/*
-	 * Load microcode on this cpu if a valid microcode is available.
-	 * This is early microcode loading procedure.
-	 */
-	load_ucode_ap();
+	if (cpu)
+		load_ucode_ap();
 
 	t = &per_cpu(cpu_tss, cpu);
 	oist = &per_cpu(orig_ist, cpu);
diff --git a/arch/x86/kernel/cpu/microcode/Makefile b/arch/x86/kernel/cpu/microcode/Makefile
index 220b1a508513..ba12e8aa4a45 100644
--- a/arch/x86/kernel/cpu/microcode/Makefile
+++ b/arch/x86/kernel/cpu/microcode/Makefile
@@ -1,4 +1,4 @@
 microcode-y				:= core.o
 obj-$(CONFIG_MICROCODE)			+= microcode.o
-microcode-$(CONFIG_MICROCODE_INTEL)	+= intel.o intel_lib.o
+microcode-$(CONFIG_MICROCODE_INTEL)	+= intel.o
 microcode-$(CONFIG_MICROCODE_AMD)	+= amd.o
diff --git a/arch/x86/kernel/cpu/microcode/amd.c b/arch/x86/kernel/cpu/microcode/amd.c
index 017bda12caae..6f353bdb3a25 100644
--- a/arch/x86/kernel/cpu/microcode/amd.c
+++ b/arch/x86/kernel/cpu/microcode/amd.c
@@ -5,6 +5,7 @@
  *  CPUs and later.
  *
  *  Copyright (C) 2008-2011 Advanced Micro Devices Inc.
+ *	          2013-2016 Borislav Petkov <bp@alien8.de>
  *
  *  Author: Peter Oruba <peter.oruba@amd.com>
  *
@@ -39,64 +40,25 @@
 
 static struct equiv_cpu_entry *equiv_cpu_table;
 
-struct ucode_patch {
-	struct list_head plist;
-	void *data;
-	u32 patch_id;
-	u16 equiv_cpu;
-};
-
-static LIST_HEAD(pcache);
-
 /*
  * This points to the current valid container of microcode patches which we will
- * save from the initrd before jettisoning its contents.
+ * save from the initrd/builtin before jettisoning its contents.
  */
-static u8 *container;
-static size_t container_size;
-static bool ucode_builtin;
+struct container {
+	u8 *data;
+	size_t size;
+} cont;
 
 static u32 ucode_new_rev;
 static u8 amd_ucode_patch[PATCH_MAX_SIZE];
 static u16 this_equiv_id;
 
-static struct cpio_data ucode_cpio;
-
-static struct cpio_data __init find_ucode_in_initrd(void)
-{
-#ifdef CONFIG_BLK_DEV_INITRD
-	char *path;
-	void *start;
-	size_t size;
-
-	/*
-	 * Microcode patch container file is prepended to the initrd in cpio
-	 * format. See Documentation/x86/early-microcode.txt
-	 */
-	static __initdata char ucode_path[] = "kernel/x86/microcode/AuthenticAMD.bin";
-
-#ifdef CONFIG_X86_32
-	struct boot_params *p;
-
-	/*
-	 * On 32-bit, early load occurs before paging is turned on so we need
-	 * to use physical addresses.
-	 */
-	p       = (struct boot_params *)__pa_nodebug(&boot_params);
-	path    = (char *)__pa_nodebug(ucode_path);
-	start   = (void *)p->hdr.ramdisk_image;
-	size    = p->hdr.ramdisk_size;
-#else
-	path    = ucode_path;
-	start   = (void *)(boot_params.hdr.ramdisk_image + PAGE_OFFSET);
-	size    = boot_params.hdr.ramdisk_size;
-#endif /* !CONFIG_X86_32 */
-
-	return find_cpio_data(path, start, size, NULL);
-#else
-	return (struct cpio_data){ NULL, 0, "" };
-#endif
-}
+/*
+ * Microcode patch container file is prepended to the initrd in cpio
+ * format. See Documentation/x86/early-microcode.txt
+ */
+static const char
+ucode_path[] __maybe_unused = "kernel/x86/microcode/AuthenticAMD.bin";
 
 static size_t compute_container_size(u8 *data, u32 total_size)
 {
@@ -135,48 +97,48 @@ static size_t compute_container_size(u8 *data, u32 total_size)
 	return size;
 }
 
+static inline u16 find_equiv_id(struct equiv_cpu_entry *equiv_cpu_table,
+				unsigned int sig)
+{
+	int i = 0;
+
+	if (!equiv_cpu_table)
+		return 0;
+
+	while (equiv_cpu_table[i].installed_cpu != 0) {
+		if (sig == equiv_cpu_table[i].installed_cpu)
+			return equiv_cpu_table[i].equiv_cpu;
+
+		i++;
+	}
+	return 0;
+}
+
 /*
- * Early load occurs before we can vmalloc(). So we look for the microcode
- * patch container file in initrd, traverse equivalent cpu table, look for a
- * matching microcode patch, and update, all in initrd memory in place.
- * When vmalloc() is available for use later -- on 64-bit during first AP load,
- * and on 32-bit during save_microcode_in_initrd_amd() -- we can call
- * load_microcode_amd() to save equivalent cpu table and microcode patches in
- * kernel heap memory.
+ * This scans the ucode blob for the proper container as we can have multiple
+ * containers glued together.
  */
-static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
+static struct container
+find_proper_container(u8 *ucode, size_t size, u16 *ret_id)
 {
+	struct container ret = { NULL, 0 };
+	u32 eax, ebx, ecx, edx;
 	struct equiv_cpu_entry *eq;
-	size_t *cont_sz;
-	u32 *header;
-	u8  *data, **cont;
-	u8 (*patch)[PATCH_MAX_SIZE];
-	u16 eq_id = 0;
 	int offset, left;
-	u32 rev, eax, ebx, ecx, edx;
-	u32 *new_rev;
-
-#ifdef CONFIG_X86_32
-	new_rev = (u32 *)__pa_nodebug(&ucode_new_rev);
-	cont_sz = (size_t *)__pa_nodebug(&container_size);
-	cont	= (u8 **)__pa_nodebug(&container);
-	patch	= (u8 (*)[PATCH_MAX_SIZE])__pa_nodebug(&amd_ucode_patch);
-#else
-	new_rev = &ucode_new_rev;
-	cont_sz = &container_size;
-	cont	= &container;
-	patch	= &amd_ucode_patch;
-#endif
+	u16 eq_id = 0;
+	u32 *header;
+	u8 *data;
 
 	data   = ucode;
 	left   = size;
 	header = (u32 *)data;
 
+
 	/* find equiv cpu table */
 	if (header[0] != UCODE_MAGIC ||
 	    header[1] != UCODE_EQUIV_CPU_TABLE_TYPE || /* type */
 	    header[2] == 0)                            /* size */
-		return;
+		return ret;
 
 	eax = 0x00000001;
 	ecx = 0;
@@ -185,7 +147,7 @@ static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
 	while (left > 0) {
 		eq = (struct equiv_cpu_entry *)(data + CONTAINER_HDR_SZ);
 
-		*cont = data;
+		ret.data = data;
 
 		/* Advance past the container header */
 		offset = header[2] + CONTAINER_HDR_SZ;
@@ -194,15 +156,15 @@ static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
 
 		eq_id = find_equiv_id(eq, eax);
 		if (eq_id) {
-			this_equiv_id = eq_id;
-			*cont_sz = compute_container_size(*cont, left + offset);
+			ret.size = compute_container_size(ret.data, left + offset);
 
 			/*
 			 * truncate how much we need to iterate over in the
 			 * ucode update loop below
 			 */
-			left = *cont_sz - offset;
-			break;
+			left = ret.size - offset;
+			*ret_id = eq_id;
+			return ret;
 		}
 
 		/*
@@ -212,6 +174,7 @@ static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
 		 */
 		while (left > 0) {
 			header = (u32 *)data;
+
 			if (header[0] == UCODE_MAGIC &&
 			    header[1] == UCODE_EQUIV_CPU_TABLE_TYPE)
 				break;
@@ -226,14 +189,64 @@ static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
 		ucode     = data;
 	}
 
-	if (!eq_id) {
-		*cont = NULL;
-		*cont_sz = 0;
-		return;
-	}
+	return ret;
+}
+
+static int __apply_microcode_amd(struct microcode_amd *mc_amd)
+{
+	u32 rev, dummy;
+
+	native_wrmsrl(MSR_AMD64_PATCH_LOADER, (u64)(long)&mc_amd->hdr.data_code);
+
+	/* verify patch application was successful */
+	native_rdmsr(MSR_AMD64_PATCH_LEVEL, rev, dummy);
+	if (rev != mc_amd->hdr.patch_id)
+		return -1;
+
+	return 0;
+}
+
+/*
+ * Early load occurs before we can vmalloc(). So we look for the microcode
+ * patch container file in initrd, traverse equivalent cpu table, look for a
+ * matching microcode patch, and update, all in initrd memory in place.
+ * When vmalloc() is available for use later -- on 64-bit during first AP load,
+ * and on 32-bit during save_microcode_in_initrd_amd() -- we can call
+ * load_microcode_amd() to save equivalent cpu table and microcode patches in
+ * kernel heap memory.
+ */
+static struct container
+apply_microcode_early_amd(void *ucode, size_t size, bool save_patch)
+{
+	struct container ret = { NULL, 0 };
+	u8 (*patch)[PATCH_MAX_SIZE];
+	int offset, left;
+	u32 rev, *header;
+	u8  *data;
+	u16 eq_id = 0;
+	u32 *new_rev;
+
+#ifdef CONFIG_X86_32
+	new_rev = (u32 *)__pa_nodebug(&ucode_new_rev);
+	patch	= (u8 (*)[PATCH_MAX_SIZE])__pa_nodebug(&amd_ucode_patch);
+#else
+	new_rev = &ucode_new_rev;
+	patch	= &amd_ucode_patch;
+#endif
 
 	if (check_current_patch_level(&rev, true))
-		return;
+		return (struct container){ NULL, 0 };
+
+	ret = find_proper_container(ucode, size, &eq_id);
+	if (!eq_id)
+		return (struct container){ NULL, 0 };
+
+	this_equiv_id = eq_id;
+	header = (u32 *)ret.data;
+
+	/* We're pointing to an equiv table, skip over it. */
+	data = ret.data +  header[2] + CONTAINER_HDR_SZ;
+	left = ret.size - (header[2] + CONTAINER_HDR_SZ);
 
 	while (left > 0) {
 		struct microcode_amd *mc;
@@ -252,8 +265,7 @@ static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
 				*new_rev = rev;
 
 				if (save_patch)
-					memcpy(patch, mc,
-					       min_t(u32, header[1], PATCH_MAX_SIZE));
+					memcpy(patch, mc, min_t(u32, header[1], PATCH_MAX_SIZE));
 			}
 		}
 
@@ -261,10 +273,10 @@ static void apply_ucode_in_initrd(void *ucode, size_t size, bool save_patch)
 		data   += offset;
 		left   -= offset;
 	}
+	return ret;
 }
 
-static bool __init load_builtin_amd_microcode(struct cpio_data *cp,
-					      unsigned int family)
+static bool get_builtin_microcode(struct cpio_data *cp, unsigned int family)
 {
 #ifdef CONFIG_X86_64
 	char fw_name[36] = "amd-ucode/microcode_amd.bin";
@@ -281,47 +293,45 @@ static bool __init load_builtin_amd_microcode(struct cpio_data *cp,
 
 void __init load_ucode_amd_bsp(unsigned int family)
 {
+	struct ucode_cpu_info *uci;
 	struct cpio_data cp;
-	bool *builtin;
-	void **data;
-	size_t *size;
+	const char *path;
+	bool use_pa;
 
-#ifdef CONFIG_X86_32
-	data =  (void **)__pa_nodebug(&ucode_cpio.data);
-	size = (size_t *)__pa_nodebug(&ucode_cpio.size);
-	builtin = (bool *)__pa_nodebug(&ucode_builtin);
-#else
-	data = &ucode_cpio.data;
-	size = &ucode_cpio.size;
-	builtin = &ucode_builtin;
-#endif
+	if (IS_ENABLED(CONFIG_X86_32)) {
+		uci	= (struct ucode_cpu_info *)__pa_nodebug(ucode_cpu_info);
+		path	= (const char *)__pa_nodebug(ucode_path);
+		use_pa	= true;
+	} else {
+		uci     = ucode_cpu_info;
+		path	= ucode_path;
+		use_pa	= false;
+	}
 
-	*builtin = load_builtin_amd_microcode(&cp, family);
-	if (!*builtin)
-		cp = find_ucode_in_initrd();
+	if (!get_builtin_microcode(&cp, family))
+		cp = find_microcode_in_initrd(path, use_pa);
 
 	if (!(cp.data && cp.size))
 		return;
 
-	*data = cp.data;
-	*size = cp.size;
+	/* Get BSP's CPUID.EAX(1), needed in load_microcode_amd() */
+	uci->cpu_sig.sig = cpuid_eax(1);
 
-	apply_ucode_in_initrd(cp.data, cp.size, true);
+	apply_microcode_early_amd(cp.data, cp.size, true);
 }
 
 #ifdef CONFIG_X86_32
 /*
  * On 32-bit, since AP's early load occurs before paging is turned on, we
- * cannot traverse cpu_equiv_table and pcache in kernel heap memory. So during
- * cold boot, AP will apply_ucode_in_initrd() just like the BSP. During
- * save_microcode_in_initrd_amd() BSP's patch is copied to amd_ucode_patch,
+ * cannot traverse cpu_equiv_table and microcode_cache in kernel heap memory.
+ * So during cold boot, AP will apply_ucode_in_initrd() just like the BSP.
+ * In save_microcode_in_initrd_amd() BSP's patch is copied to amd_ucode_patch,
  * which is used upon resume from suspend.
  */
-void load_ucode_amd_ap(void)
+void load_ucode_amd_ap(unsigned int family)
 {
 	struct microcode_amd *mc;
-	size_t *usize;
-	void **ucode;
+	struct cpio_data cp;
 
 	mc = (struct microcode_amd *)__pa_nodebug(amd_ucode_patch);
 	if (mc->hdr.patch_id && mc->hdr.processor_rev_id) {
@@ -329,60 +339,63 @@ void load_ucode_amd_ap(void)
 		return;
 	}
 
-	ucode = (void *)__pa_nodebug(&container);
-	usize = (size_t *)__pa_nodebug(&container_size);
+	if (!get_builtin_microcode(&cp, family))
+		cp = find_microcode_in_initrd((const char *)__pa_nodebug(ucode_path), true);
 
-	if (!*ucode || !*usize)
+	if (!(cp.data && cp.size))
 		return;
 
-	apply_ucode_in_initrd(*ucode, *usize, false);
-}
-
-static void __init collect_cpu_sig_on_bsp(void *arg)
-{
-	unsigned int cpu = smp_processor_id();
-	struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
-
-	uci->cpu_sig.sig = cpuid_eax(0x00000001);
-}
-
-static void __init get_bsp_sig(void)
-{
-	unsigned int bsp = boot_cpu_data.cpu_index;
-	struct ucode_cpu_info *uci = ucode_cpu_info + bsp;
-
-	if (!uci->cpu_sig.sig)
-		smp_call_function_single(bsp, collect_cpu_sig_on_bsp, NULL, 1);
+	/*
+	 * This would set amd_ucode_patch above so that the following APs can
+	 * use it directly instead of going down this path again.
+	 */
+	apply_microcode_early_amd(cp.data, cp.size, true);
 }
 #else
-void load_ucode_amd_ap(void)
+void load_ucode_amd_ap(unsigned int family)
 {
-	unsigned int cpu = smp_processor_id();
 	struct equiv_cpu_entry *eq;
 	struct microcode_amd *mc;
-	u8 *cont = container;
 	u32 rev, eax;
 	u16 eq_id;
 
-	/* Exit if called on the BSP. */
-	if (!cpu)
+	/* 64-bit runs with paging enabled, thus early==false. */
+	if (check_current_patch_level(&rev, false))
 		return;
 
-	if (!container)
-		return;
+	/* First AP hasn't cached it yet, go through the blob. */
+	if (!cont.data) {
+		struct cpio_data cp = { NULL, 0, "" };
 
-	/*
-	 * 64-bit runs with paging enabled, thus early==false.
-	 */
-	if (check_current_patch_level(&rev, false))
-		return;
+		if (cont.size == -1)
+			return;
 
-	/* Add CONFIG_RANDOMIZE_MEMORY offset. */
-	if (!ucode_builtin)
-		cont += PAGE_OFFSET - __PAGE_OFFSET_BASE;
+reget:
+		if (!get_builtin_microcode(&cp, family)) {
+#ifdef CONFIG_BLK_DEV_INITRD
+			cp = find_cpio_data(ucode_path, (void *)initrd_start,
+					    initrd_end - initrd_start, NULL);
+#endif
+			if (!(cp.data && cp.size)) {
+				/*
+				 * Mark it so that other APs do not scan again
+				 * for no real reason and slow down boot
+				 * needlessly.
+				 */
+				cont.size = -1;
+				return;
+			}
+		}
+
+		cont = apply_microcode_early_amd(cp.data, cp.size, false);
+		if (!(cont.data && cont.size)) {
+			cont.size = -1;
+			return;
+		}
+	}
 
 	eax = cpuid_eax(0x00000001);
-	eq  = (struct equiv_cpu_entry *)(cont + CONTAINER_HDR_SZ);
+	eq  = (struct equiv_cpu_entry *)(cont.data + CONTAINER_HDR_SZ);
 
 	eq_id = find_equiv_id(eq, eax);
 	if (!eq_id)
@@ -397,61 +410,50 @@ void load_ucode_amd_ap(void)
 		}
 
 	} else {
-		if (!ucode_cpio.data)
-			return;
 
 		/*
 		 * AP has a different equivalence ID than BSP, looks like
 		 * mixed-steppings silicon so go through the ucode blob anew.
 		 */
-		apply_ucode_in_initrd(ucode_cpio.data, ucode_cpio.size, false);
+		goto reget;
 	}
 }
-#endif
+#endif /* CONFIG_X86_32 */
+
+static enum ucode_state
+load_microcode_amd(int cpu, u8 family, const u8 *data, size_t size);
 
-int __init save_microcode_in_initrd_amd(void)
+int __init save_microcode_in_initrd_amd(unsigned int fam)
 {
-	unsigned long cont;
-	int retval = 0;
 	enum ucode_state ret;
-	u8 *cont_va;
-	u32 eax;
+	int retval = 0;
+	u16 eq_id;
 
-	if (!container)
-		return -EINVAL;
+	if (!cont.data) {
+		if (IS_ENABLED(CONFIG_X86_32) && (cont.size != -1)) {
+			struct cpio_data cp = { NULL, 0, "" };
 
-#ifdef CONFIG_X86_32
-	get_bsp_sig();
-	cont	= (unsigned long)container;
-	cont_va = __va(container);
-#else
-	/*
-	 * We need the physical address of the container for both bitness since
-	 * boot_params.hdr.ramdisk_image is a physical address.
-	 */
-	cont    = __pa_nodebug(container);
-	cont_va = container;
+#ifdef CONFIG_BLK_DEV_INITRD
+			cp = find_cpio_data(ucode_path, (void *)initrd_start,
+					    initrd_end - initrd_start, NULL);
 #endif
 
-	/*
-	 * Take into account the fact that the ramdisk might get relocated and
-	 * therefore we need to recompute the container's position in virtual
-	 * memory space.
-	 */
-	if (relocated_ramdisk)
-		container = (u8 *)(__va(relocated_ramdisk) +
-			     (cont - boot_params.hdr.ramdisk_image));
-	else
-		container = cont_va;
+			if (!(cp.data && cp.size)) {
+				cont.size = -1;
+				return -EINVAL;
+			}
 
-	/* Add CONFIG_RANDOMIZE_MEMORY offset. */
-	if (!ucode_builtin)
-		container += PAGE_OFFSET - __PAGE_OFFSET_BASE;
+			cont = find_proper_container(cp.data, cp.size, &eq_id);
+			if (!eq_id) {
+				cont.size = -1;
+				return -EINVAL;
+			}
 
-	eax   = cpuid_eax(0x00000001);
-	eax   = ((eax >> 8) & 0xf) + ((eax >> 20) & 0xff);
+		} else
+			return -EINVAL;
+	}
 
-	ret = load_microcode_amd(smp_processor_id(), eax, container, container_size);
+	ret = load_microcode_amd(smp_processor_id(), fam, cont.data, cont.size);
 	if (ret != UCODE_OK)
 		retval = -EINVAL;
 
@@ -459,8 +461,8 @@ int __init save_microcode_in_initrd_amd(void)
 	 * This will be freed any msec now, stash patches for the current
 	 * family and switch to patch cache for cpu hotplug, etc later.
 	 */
-	container = NULL;
-	container_size = 0;
+	cont.data = NULL;
+	cont.size = 0;
 
 	return retval;
 }
@@ -478,8 +480,10 @@ void reload_ucode_amd(void)
 		return;
 
 	mc = (struct microcode_amd *)amd_ucode_patch;
+	if (!mc)
+		return;
 
-	if (mc && rev < mc->hdr.patch_id) {
+	if (rev < mc->hdr.patch_id) {
 		if (!__apply_microcode_amd(mc)) {
 			ucode_new_rev = mc->hdr.patch_id;
 			pr_info("reload patch_level=0x%08x\n", ucode_new_rev);
@@ -513,7 +517,7 @@ static struct ucode_patch *cache_find_patch(u16 equiv_cpu)
 {
 	struct ucode_patch *p;
 
-	list_for_each_entry(p, &pcache, plist)
+	list_for_each_entry(p, &microcode_cache, plist)
 		if (p->equiv_cpu == equiv_cpu)
 			return p;
 	return NULL;
@@ -523,7 +527,7 @@ static void update_cache(struct ucode_patch *new_patch)
 {
 	struct ucode_patch *p;
 
-	list_for_each_entry(p, &pcache, plist) {
+	list_for_each_entry(p, &microcode_cache, plist) {
 		if (p->equiv_cpu == new_patch->equiv_cpu) {
 			if (p->patch_id >= new_patch->patch_id)
 				/* we already have the latest patch */
@@ -536,14 +540,14 @@ static void update_cache(struct ucode_patch *new_patch)
 		}
 	}
 	/* no patch found, add it */
-	list_add_tail(&new_patch->plist, &pcache);
+	list_add_tail(&new_patch->plist, &microcode_cache);
 }
 
 static void free_cache(void)
 {
 	struct ucode_patch *p, *tmp;
 
-	list_for_each_entry_safe(p, tmp, &pcache, plist) {
+	list_for_each_entry_safe(p, tmp, &microcode_cache, plist) {
 		__list_del(p->plist.prev, p->plist.next);
 		kfree(p->data);
 		kfree(p);
@@ -663,21 +667,7 @@ bool check_current_patch_level(u32 *rev, bool early)
 	return ret;
 }
 
-int __apply_microcode_amd(struct microcode_amd *mc_amd)
-{
-	u32 rev, dummy;
-
-	native_wrmsrl(MSR_AMD64_PATCH_LOADER, (u64)(long)&mc_amd->hdr.data_code);
-
-	/* verify patch application was successful */
-	native_rdmsr(MSR_AMD64_PATCH_LEVEL, rev, dummy);
-	if (rev != mc_amd->hdr.patch_id)
-		return -1;
-
-	return 0;
-}
-
-int apply_microcode_amd(int cpu)
+static int apply_microcode_amd(int cpu)
 {
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
 	struct microcode_amd *mc_amd;
@@ -860,7 +850,8 @@ static enum ucode_state __load_microcode_amd(u8 family, const u8 *data,
 	return UCODE_OK;
 }
 
-enum ucode_state load_microcode_amd(int cpu, u8 family, const u8 *data, size_t size)
+static enum ucode_state
+load_microcode_amd(int cpu, u8 family, const u8 *data, size_t size)
 {
 	enum ucode_state ret;
 
diff --git a/arch/x86/kernel/cpu/microcode/core.c b/arch/x86/kernel/cpu/microcode/core.c
index 5ce5155f0695..6996413c78c3 100644
--- a/arch/x86/kernel/cpu/microcode/core.c
+++ b/arch/x86/kernel/cpu/microcode/core.c
@@ -3,7 +3,7 @@
  *
  * Copyright (C) 2000-2006 Tigran Aivazian <tigran@aivazian.fsnet.co.uk>
  *	      2006	Shaohua Li <shaohua.li@intel.com>
- *	      2013-2015	Borislav Petkov <bp@alien8.de>
+ *	      2013-2016	Borislav Petkov <bp@alien8.de>
  *
  * X86 CPU microcode early update for Linux:
  *
@@ -39,12 +39,15 @@
 #include <asm/microcode.h>
 #include <asm/processor.h>
 #include <asm/cmdline.h>
+#include <asm/setup.h>
 
-#define MICROCODE_VERSION	"2.01"
+#define DRIVER_VERSION	"2.2"
 
 static struct microcode_ops	*microcode_ops;
 static bool dis_ucode_ldr;
 
+LIST_HEAD(microcode_cache);
+
 /*
  * Synchronization.
  *
@@ -167,7 +170,7 @@ void load_ucode_ap(void)
 		break;
 	case X86_VENDOR_AMD:
 		if (family >= 0x10)
-			load_ucode_amd_ap();
+			load_ucode_amd_ap(family);
 		break;
 	default:
 		break;
@@ -185,7 +188,7 @@ static int __init save_microcode_in_initrd(void)
 		break;
 	case X86_VENDOR_AMD:
 		if (c->x86 >= 0x10)
-			return save_microcode_in_initrd_amd();
+			return save_microcode_in_initrd_amd(c->x86);
 		break;
 	default:
 		break;
@@ -194,6 +197,58 @@ static int __init save_microcode_in_initrd(void)
 	return -EINVAL;
 }
 
+struct cpio_data find_microcode_in_initrd(const char *path, bool use_pa)
+{
+#ifdef CONFIG_BLK_DEV_INITRD
+	unsigned long start = 0;
+	size_t size;
+
+#ifdef CONFIG_X86_32
+	struct boot_params *params;
+
+	if (use_pa)
+		params = (struct boot_params *)__pa_nodebug(&boot_params);
+	else
+		params = &boot_params;
+
+	size = params->hdr.ramdisk_size;
+
+	/*
+	 * Set start only if we have an initrd image. We cannot use initrd_start
+	 * because it is not set that early yet.
+	 */
+	if (size)
+		start = params->hdr.ramdisk_image;
+
+# else /* CONFIG_X86_64 */
+	size  = (unsigned long)boot_params.ext_ramdisk_size << 32;
+	size |= boot_params.hdr.ramdisk_size;
+
+	if (size) {
+		start  = (unsigned long)boot_params.ext_ramdisk_image << 32;
+		start |= boot_params.hdr.ramdisk_image;
+
+		start += PAGE_OFFSET;
+	}
+# endif
+
+	/*
+	 * Did we relocate the ramdisk?
+	 *
+	 * So we possibly relocate the ramdisk *after* applying microcode on the
+	 * BSP so we rely on use_pa (use physical addresses) - even if it is not
+	 * absolutely correct - to determine whether we've done the ramdisk
+	 * relocation already.
+	 */
+	if (!use_pa && relocated_ramdisk)
+		start = initrd_start;
+
+	return find_cpio_data(path, (void *)start, size, NULL);
+#else /* !CONFIG_BLK_DEV_INITRD */
+	return (struct cpio_data){ NULL, 0, "" };
+#endif
+}
+
 void reload_early_microcode(void)
 {
 	int vendor, family;
@@ -453,16 +508,17 @@ static struct attribute_group mc_attr_group = {
 
 static void microcode_fini_cpu(int cpu)
 {
-	microcode_ops->microcode_fini_cpu(cpu);
+	if (microcode_ops->microcode_fini_cpu)
+		microcode_ops->microcode_fini_cpu(cpu);
 }
 
 static enum ucode_state microcode_resume_cpu(int cpu)
 {
-	pr_debug("CPU%d updated upon resume\n", cpu);
-
 	if (apply_microcode_on_target(cpu))
 		return UCODE_ERROR;
 
+	pr_debug("CPU%d updated upon resume\n", cpu);
+
 	return UCODE_OK;
 }
 
@@ -496,6 +552,9 @@ static enum ucode_state microcode_update_cpu(int cpu)
 {
 	struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
 
+	/* Refresh CPU microcode revision after resume. */
+	collect_cpu_info(cpu);
+
 	if (uci->valid)
 		return microcode_resume_cpu(cpu);
 
@@ -579,12 +638,7 @@ static int mc_cpu_down_prep(unsigned int cpu)
 	/* Suspend is in progress, only remove the interface */
 	sysfs_remove_group(&dev->kobj, &mc_attr_group);
 	pr_debug("CPU%d removed\n", cpu);
-	/*
-	 * When a CPU goes offline, don't free up or invalidate the copy of
-	 * the microcode in kernel memory, so that we can reuse it when the
-	 * CPU comes back online without unnecessarily requesting the userspace
-	 * for it again.
-	 */
+
 	return 0;
 }
 
@@ -649,8 +703,7 @@ int __init microcode_init(void)
 	cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN, "x86/microcode:online",
 				  mc_cpu_online, mc_cpu_down_prep);
 
-	pr_info("Microcode Update Driver: v" MICROCODE_VERSION
-		" <tigran@aivazian.fsnet.co.uk>, Peter Oruba\n");
+	pr_info("Microcode Update Driver: v%s.", DRIVER_VERSION);
 
 	return 0;
 
diff --git a/arch/x86/kernel/cpu/microcode/intel.c b/arch/x86/kernel/cpu/microcode/intel.c
index cdc0deab00c9..54d50c3694d8 100644
--- a/arch/x86/kernel/cpu/microcode/intel.c
+++ b/arch/x86/kernel/cpu/microcode/intel.c
@@ -39,125 +39,83 @@
 #include <asm/setup.h>
 #include <asm/msr.h>
 
-/*
- * Temporary microcode blobs pointers storage. We note here during early load
- * the pointers to microcode blobs we've got from whatever storage (detached
- * initrd, builtin). Later on, we put those into final storage
- * mc_saved_data.mc_saved.
- *
- * Important: those are offsets from the beginning of initrd or absolute
- * addresses within the kernel image when built-in.
- */
-static unsigned long mc_tmp_ptrs[MAX_UCODE_COUNT];
-
-static struct mc_saved_data {
-	unsigned int num_saved;
-	struct microcode_intel **mc_saved;
-} mc_saved_data;
+static const char ucode_path[] = "kernel/x86/microcode/GenuineIntel.bin";
 
-/* Microcode blobs within the initrd. 0 if builtin. */
-static struct ucode_blobs {
-	unsigned long start;
-	bool valid;
-} blobs;
+/* Current microcode patch used in early patching */
+struct microcode_intel *intel_ucode_patch;
 
-/* Go through saved patches and find the one suitable for the current CPU. */
-static enum ucode_state
-find_microcode_patch(struct microcode_intel **saved,
-		     unsigned int num_saved, struct ucode_cpu_info *uci)
+static inline bool cpu_signatures_match(unsigned int s1, unsigned int p1,
+					unsigned int s2, unsigned int p2)
 {
-	struct microcode_intel *ucode_ptr, *new_mc = NULL;
-	struct microcode_header_intel *mc_hdr;
-	int new_rev, ret, i;
-
-	new_rev = uci->cpu_sig.rev;
-
-	for (i = 0; i < num_saved; i++) {
-		ucode_ptr = saved[i];
-		mc_hdr	  = (struct microcode_header_intel *)ucode_ptr;
-
-		ret = has_newer_microcode(ucode_ptr,
-					  uci->cpu_sig.sig,
-					  uci->cpu_sig.pf,
-					  new_rev);
-		if (!ret)
-			continue;
-
-		new_rev = mc_hdr->rev;
-		new_mc  = ucode_ptr;
-	}
+	if (s1 != s2)
+		return false;