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

Pull x86 microcode loading update from Ingo Molnar:
 "Two main changes that improve microcode loading on AMD CPUs:

   - Add support for all-in-one binary microcode files that concatenate
     the microcode images of multiple processor families, by Jacob Shin

   - Add early microcode loading (embedded in the initrd) support, also
     by Jacob Shin"

* 'x86-microcode-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, microcode, amd: Another early loading fixup
  x86, microcode, amd: Allow multiple families' bin files appended together
  x86, microcode, amd: Make find_ucode_in_initrd() __init
  x86, microcode, amd: Fix warnings and errors on with CONFIG_MICROCODE=m
  x86, microcode, amd: Early microcode patch loading support for AMD
  x86, microcode, amd: Refactor functions to prepare for early loading
  x86, microcode: Vendor abstract out save_microcode_in_initrd()
  x86, microcode, intel: Correct typo in printk

5 files changed, 415 insertions, 76 deletions

diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index 4ce822e..612ed34 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -93,6 +93,7 @@ obj-$(CONFIG_MICROCODE_INTEL_LIB)	+= microcode_intel_lib.o
 microcode-y				:= microcode_core.o
 microcode-$(CONFIG_MICROCODE_INTEL)	+= microcode_intel.o
 microcode-$(CONFIG_MICROCODE_AMD)	+= microcode_amd.o
+obj-$(CONFIG_MICROCODE_AMD_EARLY)	+= microcode_amd_early.o
 obj-$(CONFIG_MICROCODE)			+= microcode.o
 
 obj-$(CONFIG_X86_CHECK_BIOS_CORRUPTION) += check.o
diff --git a/arch/x86/kernel/microcode_amd.c b/arch/x86/kernel/microcode_amd.c
index efdec7c..47ebb1d 100644
--- a/arch/x86/kernel/microcode_amd.c
+++ b/arch/x86/kernel/microcode_amd.c
@@ -31,48 +31,12 @@
 #include <asm/microcode.h>
 #include <asm/processor.h>
 #include <asm/msr.h>
+#include <asm/microcode_amd.h>
 
 MODULE_DESCRIPTION("AMD Microcode Update Driver");
 MODULE_AUTHOR("Peter Oruba");
 MODULE_LICENSE("GPL v2");
 
-#define UCODE_MAGIC                0x00414d44
-#define UCODE_EQUIV_CPU_TABLE_TYPE 0x00000000
-#define UCODE_UCODE_TYPE           0x00000001
-
-struct equiv_cpu_entry {
-	u32	installed_cpu;
-	u32	fixed_errata_mask;
-	u32	fixed_errata_compare;
-	u16	equiv_cpu;
-	u16	res;
-} __attribute__((packed));
-
-struct microcode_header_amd {
-	u32	data_code;
-	u32	patch_id;
-	u16	mc_patch_data_id;
-	u8	mc_patch_data_len;
-	u8	init_flag;
-	u32	mc_patch_data_checksum;
-	u32	nb_dev_id;
-	u32	sb_dev_id;
-	u16	processor_rev_id;
-	u8	nb_rev_id;
-	u8	sb_rev_id;
-	u8	bios_api_rev;
-	u8	reserved1[3];
-	u32	match_reg[8];
-} __attribute__((packed));
-
-struct microcode_amd {
-	struct microcode_header_amd	hdr;
-	unsigned int			mpb[0];
-};
-
-#define SECTION_HDR_SIZE	8
-#define CONTAINER_HDR_SZ	12
-
 static struct equiv_cpu_entry *equiv_cpu_table;
 
 struct ucode_patch {
@@ -84,21 +48,10 @@ struct ucode_patch {
 
 static LIST_HEAD(pcache);
 
-static u16 find_equiv_id(unsigned int cpu)
+static u16 __find_equiv_id(unsigned int cpu)
 {
 	struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
-	int i = 0;
-
-	if (!equiv_cpu_table)
-		return 0;
-
-	while (equiv_cpu_table[i].installed_cpu != 0) {
-		if (uci->cpu_sig.sig == equiv_cpu_table[i].installed_cpu)
-			return equiv_cpu_table[i].equiv_cpu;
-
-		i++;
-	}
-	return 0;
+	return find_equiv_id(equiv_cpu_table, uci->cpu_sig.sig);
 }
 
 static u32 find_cpu_family_by_equiv_cpu(u16 equiv_cpu)
@@ -163,7 +116,7 @@ static struct ucode_patch *find_patch(unsigned int cpu)
 {
 	u16 equiv_id;
 
-	equiv_id = find_equiv_id(cpu);
+	equiv_id = __find_equiv_id(cpu);
 	if (!equiv_id)
 		return NULL;
 
@@ -173,9 +126,20 @@ static struct ucode_patch *find_patch(unsigned int cpu)
 static int collect_cpu_info_amd(int cpu, struct cpu_signature *csig)
 {
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
+	struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
+	struct ucode_patch *p;
 
 	csig->sig = cpuid_eax(0x00000001);
 	csig->rev = c->microcode;
+
+	/*
+	 * a patch could have been loaded early, set uci->mc so that
+	 * mc_bp_resume() can call apply_microcode()
+	 */
+	p = find_patch(cpu);
+	if (p && (p->patch_id == csig->rev))
+		uci->mc = p->data;
+
 	pr_info("CPU%d: patch_level=0x%08x\n", cpu, csig->rev);
 
 	return 0;
@@ -215,7 +179,21 @@ static unsigned int verify_patch_size(int cpu, u32 patch_size,
 	return patch_size;
 }
 
-static int apply_microcode_amd(int cpu)
+int __apply_microcode_amd(struct microcode_amd *mc_amd)
+{
+	u32 rev, dummy;
+
+	wrmsrl(MSR_AMD64_PATCH_LOADER, (u64)(long)&mc_amd->hdr.data_code);
+
+	/* verify patch application was successful */
+	rdmsr(MSR_AMD64_PATCH_LEVEL, rev, dummy);
+	if (rev != mc_amd->hdr.patch_id)
+		return -1;
+
+	return 0;
+}
+
+int apply_microcode_amd(int cpu)
 {
 	struct cpuinfo_x86 *c = &cpu_data(cpu);
 	struct microcode_amd *mc_amd;
@@ -242,19 +220,15 @@ static int apply_microcode_amd(int cpu)
 		return 0;
 	}
 
-	wrmsrl(MSR_AMD64_PATCH_LOADER, (u64)(long)&mc_amd->hdr.data_code);
-
-	/* verify patch application was successful */
-	rdmsr(MSR_AMD64_PATCH_LEVEL, rev, dummy);
-	if (rev != mc_amd->hdr.patch_id) {
+	if (__apply_microcode_amd(mc_amd))
 		pr_err("CPU%d: update failed for patch_level=0x%08x\n",
-		       cpu, mc_amd->hdr.patch_id);
-		return -1;
-	}
+			cpu, mc_amd->hdr.patch_id);
+	else
+		pr_info("CPU%d: new patch_level=0x%08x\n", cpu,
+			mc_amd->hdr.patch_id);
 
-	pr_info("CPU%d: new patch_level=0x%08x\n", cpu, rev);
-	uci->cpu_sig.rev = rev;
-	c->microcode = rev;
+	uci->cpu_sig.rev = mc_amd->hdr.patch_id;
+	c->microcode = mc_amd->hdr.patch_id;
 
 	return 0;
 }
@@ -364,7 +338,7 @@ static int verify_and_add_patch(unsigned int cpu, u8 *fw, unsigned int leftover)
 	return crnt_size;
 }
 
-static enum ucode_state load_microcode_amd(int cpu, const u8 *data, size_t size)
+static enum ucode_state __load_microcode_amd(int cpu, const u8 *data, size_t size)
 {
 	enum ucode_state ret = UCODE_ERROR;
 	unsigned int leftover;
@@ -398,6 +372,32 @@ static enum ucode_state load_microcode_amd(int cpu, const u8 *data, size_t size)
 	return UCODE_OK;
 }
 
+enum ucode_state load_microcode_amd(int cpu, const u8 *data, size_t size)
+{
+	enum ucode_state ret;
+
+	/* free old equiv table */
+	free_equiv_cpu_table();
+
+	ret = __load_microcode_amd(cpu, data, size);
+
+	if (ret != UCODE_OK)
+		cleanup();
+
+#if defined(CONFIG_MICROCODE_AMD_EARLY) && defined(CONFIG_X86_32)
+	/* save BSP's matching patch for early load */
+	if (cpu_data(cpu).cpu_index == boot_cpu_data.cpu_index) {
+		struct ucode_patch *p = find_patch(cpu);
+		if (p) {
+			memset(amd_bsp_mpb, 0, MPB_MAX_SIZE);
+			memcpy(amd_bsp_mpb, p->data, min_t(u32, ksize(p->data),
+							   MPB_MAX_SIZE));
+		}
+	}
+#endif
+	return ret;
+}
+
 /*
  * AMD microcode firmware naming convention, up to family 15h they are in
  * the legacy file:
@@ -440,12 +440,7 @@ static enum ucode_state request_microcode_amd(int cpu, struct device *device,
 		goto fw_release;
 	}
 
-	/* free old equiv table */
-	free_equiv_cpu_table();
-
 	ret = load_microcode_amd(cpu, fw->data, fw->size);
-	if (ret != UCODE_OK)
-		cleanup();
 
  fw_release:
 	release_firmware(fw);
diff --git a/arch/x86/kernel/microcode_amd_early.c b/arch/x86/kernel/microcode_amd_early.c
new file mode 100644
index 0000000..1ac6e9a
--- /dev/null
+++ b/arch/x86/kernel/microcode_amd_early.c
@@ -0,0 +1,302 @@
+/*
+ * Copyright (C) 2013 Advanced Micro Devices, Inc.
+ *
+ * Author: Jacob Shin <jacob.shin@amd.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/earlycpio.h>
+#include <linux/initrd.h>
+
+#include <asm/cpu.h>
+#include <asm/setup.h>
+#include <asm/microcode_amd.h>
+
+static bool ucode_loaded;
+static u32 ucode_new_rev;
+static unsigned long ucode_offset;
+static size_t ucode_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";
+
+static struct cpio_data __init find_ucode_in_initrd(void)
+{
+	long offset = 0;
+	char *path;
+	void *start;
+	size_t size;
+	unsigned long *uoffset;
+	size_t *usize;
+	struct cpio_data cd;
+
+#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;
+	uoffset = (unsigned long *)__pa_nodebug(&ucode_offset);
+	usize   = (size_t *)__pa_nodebug(&ucode_size);
+#else
+	path    = ucode_path;
+	start   = (void *)(boot_params.hdr.ramdisk_image + PAGE_OFFSET);
+	size    = boot_params.hdr.ramdisk_size;
+	uoffset = &ucode_offset;
+	usize   = &ucode_size;
+#endif
+
+	cd = find_cpio_data(path, start, size, &offset);
+	if (!cd.data)
+		return cd;
+
+	if (*(u32 *)cd.data != UCODE_MAGIC) {
+		cd.data = NULL;
+		cd.size = 0;
+		return cd;
+	}
+
+	*uoffset = (u8 *)cd.data - (u8 *)start;
+	*usize   = cd.size;
+
+	return cd;
+}
+
+/*
+ * 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 void __cpuinit apply_ucode_in_initrd(void *ucode, size_t size)
+{
+	struct equiv_cpu_entry *eq;
+	u32 *header;
+	u8  *data;
+	u16 eq_id = 0;
+	int offset, left;
+	u32 rev, eax;
+	u32 *new_rev;
+	unsigned long *uoffset;
+	size_t *usize;
+
+#ifdef CONFIG_X86_32
+	new_rev = (u32 *)__pa_nodebug(&ucode_new_rev);
+	uoffset = (unsigned long *)__pa_nodebug(&ucode_offset);
+	usize   = (size_t *)__pa_nodebug(&ucode_size);
+#else
+	new_rev = &ucode_new_rev;
+	uoffset = &ucode_offset;
+	usize   = &ucode_size;
+#endif
+
+	data   = ucode;
+	left   = size;
+	header = (u32 *)data;
+
+	/* find equiv cpu table */
+
+	if (header[1] != UCODE_EQUIV_CPU_TABLE_TYPE || /* type */
+	    header[2] == 0)                            /* size */
+		return;
+
+	eax = cpuid_eax(0x00000001);
+
+	while (left > 0) {
+		eq = (struct equiv_cpu_entry *)(data + CONTAINER_HDR_SZ);
+
+		offset = header[2] + CONTAINER_HDR_SZ;
+		data  += offset;
+		left  -= offset;
+
+		eq_id = find_equiv_id(eq, eax);
+		if (eq_id)
+			break;
+
+		/*
+		 * support multiple container files appended together. if this
+		 * one does not have a matching equivalent cpu entry, we fast
+		 * forward to the next container file.
+		 */
+		while (left > 0) {
+			header = (u32 *)data;
+			if (header[0] == UCODE_MAGIC &&
+			    header[1] == UCODE_EQUIV_CPU_TABLE_TYPE)
+				break;
+
+			offset = header[1] + SECTION_HDR_SIZE;
+			data  += offset;
+			left  -= offset;
+		}
+
+		/* mark where the next microcode container file starts */
+		offset    = data - (u8 *)ucode;
+		*uoffset += offset;
+		*usize   -= offset;
+		ucode     = data;
+	}
+
+	if (!eq_id) {
+		*usize = 0;
+		return;
+	}
+
+	/* find ucode and update if needed */
+
+	rdmsr(MSR_AMD64_PATCH_LEVEL, rev, eax);
+
+	while (left > 0) {
+		struct microcode_amd *mc;
+
+		header = (u32 *)data;
+		if (header[0] != UCODE_UCODE_TYPE || /* type */
+		    header[1] == 0)                  /* size */
+			break;
+
+		mc = (struct microcode_amd *)(data + SECTION_HDR_SIZE);
+		if (eq_id == mc->hdr.processor_rev_id && rev < mc->hdr.patch_id)
+			if (__apply_microcode_amd(mc) == 0) {
+				rev = mc->hdr.patch_id;
+				*new_rev = rev;
+			}
+
+		offset  = header[1] + SECTION_HDR_SIZE;
+		data   += offset;
+		left   -= offset;
+	}
+
+	/* mark where this microcode container file ends */
+	offset  = *usize - (data - (u8 *)ucode);
+	*usize -= offset;
+
+	if (!(*new_rev))
+		*usize = 0;
+}
+
+void __init load_ucode_amd_bsp(void)
+{
+	struct cpio_data cd = find_ucode_in_initrd();
+	if (!cd.data)
+		return;
+
+	apply_ucode_in_initrd(cd.data, cd.size);
+}
+
+#ifdef CONFIG_X86_32
+u8 amd_bsp_mpb[MPB_MAX_SIZE];
+
+/*
+ * 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_bsp_mpb, which
+ * is used upon resume from suspend.
+ */
+void __cpuinit load_ucode_amd_ap(void)
+{
+	struct microcode_amd *mc;
+	unsigned long *initrd;
+	unsigned long *uoffset;
+	size_t *usize;
+	void *ucode;
+
+	mc = (struct microcode_amd *)__pa(amd_bsp_mpb);
+	if (mc->hdr.patch_id && mc->hdr.processor_rev_id) {
+		__apply_microcode_amd(mc);
+		return;
+	}
+
+	initrd  = (unsigned long *)__pa(&initrd_start);
+	uoffset = (unsigned long *)__pa(&ucode_offset);
+	usize   = (size_t *)__pa(&ucode_size);
+
+	if (!*usize || !*initrd)
+		return;
+
+	ucode = (void *)((unsigned long)__pa(*initrd) + *uoffset);
+	apply_ucode_in_initrd(ucode, *usize);
+}
+
+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);
+}
+#else
+static void __cpuinit collect_cpu_info_amd_early(struct cpuinfo_x86 *c,
+						 struct ucode_cpu_info *uci)
+{
+	u32 rev, eax;
+
+	rdmsr(MSR_AMD64_PATCH_LEVEL, rev, eax);
+	eax = cpuid_eax(0x00000001);
+
+	uci->cpu_sig.sig = eax;
+	uci->cpu_sig.rev = rev;
+	c->microcode = rev;
+	c->x86 = ((eax >> 8) & 0xf) + ((eax >> 20) & 0xff);
+}
+
+void __cpuinit load_ucode_amd_ap(void)
+{
+	unsigned int cpu = smp_processor_id();
+
+	collect_cpu_info_amd_early(&cpu_data(cpu), ucode_cpu_info + cpu);
+
+	if (cpu && !ucode_loaded) {
+		void *ucode;
+
+		if (!ucode_size || !initrd_start)
+			return;
+
+		ucode = (void *)(initrd_start + ucode_offset);
+		if (load_microcode_amd(0, ucode, ucode_size) != UCODE_OK)
+			return;
+		ucode_loaded = true;
+	}
+
+	apply_microcode_amd(cpu);
+}
+#endif
+
+int __init save_microcode_in_initrd_amd(void)
+{
+	enum ucode_state ret;
+	void *ucode;
+#ifdef CONFIG_X86_32
+	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);
+#endif
+	if (ucode_new_rev)
+		pr_info("microcode: updated early to new patch_level=0x%08x\n",
+			ucode_new_rev);
+
+	if (ucode_loaded || !ucode_size || !initrd_start)
+		return 0;
+
+	ucode = (void *)(initrd_start + ucode_offset);
+	ret = load_microcode_amd(0, ucode, ucode_size);
+	if (ret != UCODE_OK)
+		return -EINVAL;
+
+	ucode_loaded = true;
+	return 0;
+}
diff --git a/arch/x86/kernel/microcode_core_early.c b/arch/x86/kernel/microcode_core_early.c
index 833d51d..86119f6 100644
--- a/arch/x86/kernel/microcode_core_early.c
+++ b/arch/x86/kernel/microcode_core_early.c
@@ -18,6 +18,7 @@
  */
 #include <linux/module.h>
 #include <asm/microcode_intel.h>
+#include <asm/microcode_amd.h>
 #include <asm/processor.h>
 
 #define QCHAR(a, b, c, d) ((a) + ((b) << 8) + ((c) << 16) + ((d) << 24))
@@ -81,8 +82,18 @@ void __init load_ucode_bsp(void)
 	vendor = x86_vendor();
 	x86 = x86_family();
 
-	if (vendor == X86_VENDOR_INTEL && x86 >= 6)
-		load_ucode_intel_bsp();
+	switch (vendor) {
+	case X86_VENDOR_INTEL:
+		if (x86 >= 6)
+			load_ucode_intel_bsp();
+		break;
+	case X86_VENDOR_AMD:
+		if (x86 >= 0x10)
+			load_ucode_amd_bsp();
+		break;
+	default:
+		break;
+	}
 }
 
 void __cpuinit load_ucode_ap(void)
@@ -95,6 +106,36 @@ void __cpuinit load_ucode_ap(void)
 	vendor = x86_vendor();
 	x86 = x86_family();
 
-	if (vendor == X86_VENDOR_INTEL && x86 >= 6)
-		load_ucode_intel_ap();
+	switch (vendor) {
+	case X86_VENDOR_INTEL:
+		if (x86 >= 6)
+			load_ucode_intel_ap();
+		break;
+	case X86_VENDOR_AMD:
+		if (x86 >= 0x10)
+			load_ucode_amd_ap();
+		break;
+	default:
+		break;
+	}
+}
+
+int __init save_microcode_in_initrd(void)
+{
+	struct cpuinfo_x86 *c = &boot_cpu_data;
+
+	switch (c->x86_vendor) {
+	case X86_VENDOR_INTEL:
+		if (c->x86 >= 6)
+			save_microcode_in_initrd_intel();
+		break;
+	case X86_VENDOR_AMD:
+		if (c->x86 >= 0x10)
+			save_microcode_in_initrd_amd();
+		break;
+	default:
+		break;
+	}
+
+	return 0;
 }
diff --git a/arch/x86/kernel/microcode_intel_early.c b/arch/x86/kernel/microcode_intel_early.c
index 2e9e128..dabef95 100644
--- a/arch/x86/kernel/microcode_intel_early.c
+++ b/arch/x86/kernel/microcode_intel_early.c
@@ -529,7 +529,7 @@ int save_mc_for_early(u8 *mc)
 	 */
 	ret = save_microcode(&mc_saved_data, mc_saved_tmp, mc_saved_count);
 	if (ret) {
-		pr_err("Can not save microcode patch.\n");
+		pr_err("Cannot save microcode patch.\n");
 		goto out;
 	}
 
@@ -699,7 +699,7 @@ static int __cpuinit apply_microcode_early(struct mc_saved_data *mc_saved_data,
  * This function converts microcode patch offsets previously stored in
  * mc_saved_in_initrd to pointers and stores the pointers in mc_saved_data.
  */
-int __init save_microcode_in_initrd(void)
+int __init save_microcode_in_initrd_intel(void)
 {
 	unsigned int count = mc_saved_data.mc_saved_count;
 	struct microcode_intel *mc_saved[MAX_UCODE_COUNT];
@@ -711,7 +711,7 @@ int __init save_microcode_in_initrd(void)
 	microcode_pointer(mc_saved, mc_saved_in_initrd, initrd_start, count);
 	ret = save_microcode(&mc_saved_data, mc_saved, count);
 	if (ret)
-		pr_err("Can not save microcod patches from initrd");
+		pr_err("Cannot save microcode patches from initrd.\n");
 
 	show_saved_mc();