Resguardo inicial

linux-7.2.3/arch/x86/kernel/cpu/mtrr/Makefile

@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0-only
+obj-y		:= mtrr.o if.o generic.o cleanup.o
+obj-$(CONFIG_X86_32) += amd.o cyrix.o centaur.o legacy.o
+

linux-7.2.3/arch/x86/kernel/cpu/mtrr/amd.c

@@ -0,0 +1,119 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/init.h>
+#include <linux/mm.h>
+#include <asm/mtrr.h>
+#include <asm/msr.h>
+
+#include "mtrr.h"
+
+static void
+amd_get_mtrr(unsigned int reg, unsigned long *base,
+	     unsigned long *size, mtrr_type *type)
+{
+	unsigned long low, high;
+
+	rdmsr(MSR_K6_UWCCR, low, high);
+	/* Upper dword is region 1, lower is region 0 */
+	if (reg == 1)
+		low = high;
+	/* The base masks off on the right alignment */
+	*base = (low & 0xFFFE0000) >> PAGE_SHIFT;
+	*type = 0;
+	if (low & 1)
+		*type = MTRR_TYPE_UNCACHABLE;
+	if (low & 2)
+		*type = MTRR_TYPE_WRCOMB;
+	if (!(low & 3)) {
+		*size = 0;
+		return;
+	}
+	/*
+	 * This needs a little explaining. The size is stored as an
+	 * inverted mask of bits of 128K granularity 15 bits long offset
+	 * 2 bits.
+	 *
+	 * So to get a size we do invert the mask and add 1 to the lowest
+	 * mask bit (4 as its 2 bits in). This gives us a size we then shift
+	 * to turn into 128K blocks.
+	 *
+	 * eg              111 1111 1111 1100      is 512K
+	 *
+	 * invert          000 0000 0000 0011
+	 * +1              000 0000 0000 0100
+	 * *128K   ...
+	 */
+	low = (~low) & 0x1FFFC;
+	*size = (low + 4) << (15 - PAGE_SHIFT);
+}
+
+/**
+ * amd_set_mtrr - Set variable MTRR register on the local CPU.
+ *
+ * @reg The register to set.
+ * @base The base address of the region.
+ * @size The size of the region. If this is 0 the region is disabled.
+ * @type The type of the region.
+ *
+ * Returns nothing.
+ */
+static void
+amd_set_mtrr(unsigned int reg, unsigned long base, unsigned long size, mtrr_type type)
+{
+	u32 regs[2];
+
+	/*
+	 * Low is MTRR0, High MTRR 1
+	 */
+	rdmsr(MSR_K6_UWCCR, regs[0], regs[1]);
+	/*
+	 * Blank to disable
+	 */
+	if (size == 0) {
+		regs[reg] = 0;
+	} else {
+		/*
+		 * Set the register to the base, the type (off by one) and an
+		 * inverted bitmask of the size The size is the only odd
+		 * bit. We are fed say 512K We invert this and we get 111 1111
+		 * 1111 1011 but if you subtract one and invert you get the
+		 * desired 111 1111 1111 1100 mask
+		 *
+		 *  But ~(x - 1) == ~x + 1 == -x. Two's complement rocks!
+		 */
+		regs[reg] = (-size >> (15 - PAGE_SHIFT) & 0x0001FFFC)
+		    | (base << PAGE_SHIFT) | (type + 1);
+	}
+
+	/*
+	 * The writeback rule is quite specific. See the manual. Its
+	 * disable local interrupts, write back the cache, set the mtrr
+	 */
+	wbinvd();
+	wrmsr(MSR_K6_UWCCR, regs[0], regs[1]);
+}
+
+static int
+amd_validate_add_page(unsigned long base, unsigned long size, unsigned int type)
+{
+	/*
+	 * Apply the K6 block alignment and size rules
+	 * In order
+	 * o Uncached or gathering only
+	 * o 128K or bigger block
+	 * o Power of 2 block
+	 * o base suitably aligned to the power
+	 */
+	if (type > MTRR_TYPE_WRCOMB || size < (1 << (17 - PAGE_SHIFT))
+	    || (size & ~(size - 1)) - size || (base & (size - 1)))
+		return -EINVAL;
+	return 0;
+}
+
+const struct mtrr_ops amd_mtrr_ops = {
+	.var_regs          = 2,
+	.set               = amd_set_mtrr,
+	.get               = amd_get_mtrr,
+	.get_free_region   = generic_get_free_region,
+	.validate_add_page = amd_validate_add_page,
+	.have_wrcomb       = positive_have_wrcomb,
+};

linux-7.2.3/arch/x86/kernel/cpu/mtrr/centaur.c

@@ -0,0 +1,112 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/init.h>
+#include <linux/mm.h>
+
+#include <asm/mtrr.h>
+#include <asm/msr.h>
+
+#include "mtrr.h"
+
+static struct {
+	unsigned long high;
+	unsigned long low;
+} centaur_mcr[8];
+
+static u8 centaur_mcr_reserved;
+static u8 centaur_mcr_type;	/* 0 for winchip, 1 for winchip2 */
+
+/**
+ * centaur_get_free_region - Get a free MTRR.
+ *
+ * @base: The starting (base) address of the region.
+ * @size: The size (in bytes) of the region.
+ *
+ * Returns: the index of the region on success, else -1 on error.
+ */
+static int
+centaur_get_free_region(unsigned long base, unsigned long size, int replace_reg)
+{
+	unsigned long lbase, lsize;
+	mtrr_type ltype;
+	int i, max;
+
+	max = num_var_ranges;
+	if (replace_reg >= 0 && replace_reg < max)
+		return replace_reg;
+
+	for (i = 0; i < max; ++i) {
+		if (centaur_mcr_reserved & (1 << i))
+			continue;
+		mtrr_if->get(i, &lbase, &lsize, &ltype);
+		if (lsize == 0)
+			return i;
+	}
+
+	return -ENOSPC;
+}
+
+static void
+centaur_get_mcr(unsigned int reg, unsigned long *base,
+		unsigned long *size, mtrr_type * type)
+{
+	*base = centaur_mcr[reg].high >> PAGE_SHIFT;
+	*size = -(centaur_mcr[reg].low & 0xfffff000) >> PAGE_SHIFT;
+	*type = MTRR_TYPE_WRCOMB;		/* write-combining  */
+
+	if (centaur_mcr_type == 1 && ((centaur_mcr[reg].low & 31) & 2))
+		*type = MTRR_TYPE_UNCACHABLE;
+	if (centaur_mcr_type == 1 && (centaur_mcr[reg].low & 31) == 25)
+		*type = MTRR_TYPE_WRBACK;
+	if (centaur_mcr_type == 0 && (centaur_mcr[reg].low & 31) == 31)
+		*type = MTRR_TYPE_WRBACK;
+}
+
+static void
+centaur_set_mcr(unsigned int reg, unsigned long base,
+		unsigned long size, mtrr_type type)
+{
+	unsigned long low, high;
+
+	if (size == 0) {
+		/* Disable */
+		high = low = 0;
+	} else {
+		high = base << PAGE_SHIFT;
+		if (centaur_mcr_type == 0) {
+			/* Only support write-combining... */
+			low = -size << PAGE_SHIFT | 0x1f;
+		} else {
+			if (type == MTRR_TYPE_UNCACHABLE)
+				low = -size << PAGE_SHIFT | 0x02; /* NC */
+			else
+				low = -size << PAGE_SHIFT | 0x09; /* WWO, WC */
+		}
+	}
+	centaur_mcr[reg].high = high;
+	centaur_mcr[reg].low = low;
+	wrmsr(MSR_IDT_MCR0 + reg, low, high);
+}
+
+static int
+centaur_validate_add_page(unsigned long base, unsigned long size, unsigned int type)
+{
+	/*
+	 * FIXME: Winchip2 supports uncached
+	 */
+	if (type != MTRR_TYPE_WRCOMB &&
+	    (centaur_mcr_type == 0 || type != MTRR_TYPE_UNCACHABLE)) {
+		pr_warn("mtrr: only write-combining%s supported\n",
+			   centaur_mcr_type ? " and uncacheable are" : " is");
+		return -EINVAL;
+	}
+	return 0;
+}
+
+const struct mtrr_ops centaur_mtrr_ops = {
+	.var_regs          = 8,
+	.set               = centaur_set_mcr,
+	.get               = centaur_get_mcr,
+	.get_free_region   = centaur_get_free_region,
+	.validate_add_page = centaur_validate_add_page,
+	.have_wrcomb       = positive_have_wrcomb,
+};

linux-7.2.3/arch/x86/kernel/cpu/mtrr/cleanup.c

@@ -0,0 +1,968 @@
+// SPDX-License-Identifier: LGPL-2.0+
+/*
+ * MTRR (Memory Type Range Register) cleanup
+ *
+ *  Copyright (C) 2009 Yinghai Lu
+ */
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/smp.h>
+#include <linux/cpu.h>
+#include <linux/mutex.h>
+#include <linux/uaccess.h>
+#include <linux/kvm_para.h>
+#include <linux/range.h>
+
+#include <asm/processor.h>
+#include <asm/e820/api.h>
+#include <asm/mtrr.h>
+#include <asm/msr.h>
+
+#include "mtrr.h"
+
+struct var_mtrr_range_state {
+	unsigned long	base_pfn;
+	unsigned long	size_pfn;
+	mtrr_type	type;
+};
+
+struct var_mtrr_state {
+	unsigned long	range_startk;
+	unsigned long	range_sizek;
+	unsigned long	chunk_sizek;
+	unsigned long	gran_sizek;
+	unsigned int	reg;
+};
+
+/* Should be related to MTRR_VAR_RANGES nums */
+#define RANGE_NUM				256
+
+static struct range __initdata		range[RANGE_NUM];
+static int __initdata				nr_range;
+
+static struct var_mtrr_range_state __initdata	range_state[RANGE_NUM];
+
+#define BIOS_BUG_MSG \
+	"WARNING: BIOS bug: VAR MTRR %d contains strange UC entry under 1M, check with your system vendor!\n"
+
+static int __init
+x86_get_mtrr_mem_range(struct range *range, int nr_range,
+		       unsigned long extra_remove_base,
+		       unsigned long extra_remove_size)
+{
+	unsigned long base, size;
+	mtrr_type type;
+	int i;
+
+	for (i = 0; i < num_var_ranges; i++) {
+		type = range_state[i].type;
+		if (type != MTRR_TYPE_WRBACK)
+			continue;
+		base = range_state[i].base_pfn;
+		size = range_state[i].size_pfn;
+		nr_range = add_range_with_merge(range, RANGE_NUM, nr_range,
+						base, base + size);
+	}
+
+	Dprintk("After WB checking\n");
+	for (i = 0; i < nr_range; i++)
+		Dprintk("MTRR MAP PFN: %016llx - %016llx\n",
+			 range[i].start, range[i].end);
+
+	/* Take out UC ranges: */
+	for (i = 0; i < num_var_ranges; i++) {
+		type = range_state[i].type;
+		if (type != MTRR_TYPE_UNCACHABLE &&
+		    type != MTRR_TYPE_WRPROT)
+			continue;
+		size = range_state[i].size_pfn;
+		if (!size)
+			continue;
+		base = range_state[i].base_pfn;
+		if (base < (1<<(20-PAGE_SHIFT)) && mtrr_state.have_fixed &&
+		    (mtrr_state.enabled & MTRR_STATE_MTRR_ENABLED) &&
+		    (mtrr_state.enabled & MTRR_STATE_MTRR_FIXED_ENABLED)) {
+			/* Var MTRR contains UC entry below 1M? Skip it: */
+			pr_warn(BIOS_BUG_MSG, i);
+			if (base + size <= (1<<(20-PAGE_SHIFT)))
+				continue;
+			size -= (1<<(20-PAGE_SHIFT)) - base;
+			base = 1<<(20-PAGE_SHIFT);
+		}
+		subtract_range(range, RANGE_NUM, base, base + size);
+	}
+	if (extra_remove_size)
+		subtract_range(range, RANGE_NUM, extra_remove_base,
+				 extra_remove_base + extra_remove_size);
+
+	Dprintk("After UC checking\n");
+	for (i = 0; i < RANGE_NUM; i++) {
+		if (!range[i].end)
+			continue;
+
+		Dprintk("MTRR MAP PFN: %016llx - %016llx\n",
+			 range[i].start, range[i].end);
+	}
+
+	/* sort the ranges */
+	nr_range = clean_sort_range(range, RANGE_NUM);
+
+	Dprintk("After sorting\n");
+	for (i = 0; i < nr_range; i++)
+		Dprintk("MTRR MAP PFN: %016llx - %016llx\n",
+			range[i].start, range[i].end);
+
+	return nr_range;
+}
+
+#ifdef CONFIG_MTRR_SANITIZER
+
+static unsigned long __init sum_ranges(struct range *range, int nr_range)
+{
+	unsigned long sum = 0;
+	int i;
+
+	for (i = 0; i < nr_range; i++)
+		sum += range[i].end - range[i].start;
+
+	return sum;
+}
+
+static int enable_mtrr_cleanup __initdata =
+	CONFIG_MTRR_SANITIZER_ENABLE_DEFAULT;
+
+static int __init disable_mtrr_cleanup_setup(char *str)
+{
+	enable_mtrr_cleanup = 0;
+	return 0;
+}
+early_param("disable_mtrr_cleanup", disable_mtrr_cleanup_setup);
+
+static int __init enable_mtrr_cleanup_setup(char *str)
+{
+	enable_mtrr_cleanup = 1;
+	return 0;
+}
+early_param("enable_mtrr_cleanup", enable_mtrr_cleanup_setup);
+
+static void __init
+set_var_mtrr(unsigned int reg, unsigned long basek, unsigned long sizek,
+	     unsigned char type)
+{
+	u32 base_lo, base_hi, mask_lo, mask_hi;
+	u64 base, mask;
+
+	if (!sizek) {
+		fill_mtrr_var_range(reg, 0, 0, 0, 0);
+		return;
+	}
+
+	mask = (1ULL << boot_cpu_data.x86_phys_bits) - 1;
+	mask &= ~((((u64)sizek) << 10) - 1);
+
+	base = ((u64)basek) << 10;
+
+	base |= type;
+	mask |= 0x800;
+
+	base_lo = base & ((1ULL<<32) - 1);
+	base_hi = base >> 32;
+
+	mask_lo = mask & ((1ULL<<32) - 1);
+	mask_hi = mask >> 32;
+
+	fill_mtrr_var_range(reg, base_lo, base_hi, mask_lo, mask_hi);
+}
+
+static void __init
+save_var_mtrr(unsigned int reg, unsigned long basek, unsigned long sizek,
+	      unsigned char type)
+{
+	range_state[reg].base_pfn = basek >> (PAGE_SHIFT - 10);
+	range_state[reg].size_pfn = sizek >> (PAGE_SHIFT - 10);
+	range_state[reg].type = type;
+}
+
+static void __init set_var_mtrr_all(void)
+{
+	unsigned long basek, sizek;
+	unsigned char type;
+	unsigned int reg;
+
+	for (reg = 0; reg < num_var_ranges; reg++) {
+		basek = range_state[reg].base_pfn << (PAGE_SHIFT - 10);
+		sizek = range_state[reg].size_pfn << (PAGE_SHIFT - 10);
+		type = range_state[reg].type;
+
+		set_var_mtrr(reg, basek, sizek, type);
+	}
+}
+
+static unsigned long to_size_factor(unsigned long sizek, char *factorp)
+{
+	unsigned long base = sizek;
+	char factor;
+
+	if (base & ((1<<10) - 1)) {
+		/* Not MB-aligned: */
+		factor = 'K';
+	} else if (base & ((1<<20) - 1)) {
+		factor = 'M';
+		base >>= 10;
+	} else {
+		factor = 'G';
+		base >>= 20;
+	}
+
+	*factorp = factor;
+
+	return base;
+}
+
+static unsigned int __init
+range_to_mtrr(unsigned int reg, unsigned long range_startk,
+	      unsigned long range_sizek, unsigned char type)
+{
+	if (!range_sizek || (reg >= num_var_ranges))
+		return reg;
+
+	while (range_sizek) {
+		unsigned long max_align, align;
+		unsigned long sizek;
+
+		/* Compute the maximum size with which we can make a range: */
+		if (range_startk)
+			max_align = __ffs(range_startk);
+		else
+			max_align = BITS_PER_LONG - 1;
+
+		align = __fls(range_sizek);
+		if (align > max_align)
+			align = max_align;
+
+		sizek = 1UL << align;
+		if (mtrr_debug) {
+			char start_factor = 'K', size_factor = 'K';
+			unsigned long start_base, size_base;
+
+			start_base = to_size_factor(range_startk, &start_factor);
+			size_base = to_size_factor(sizek, &size_factor);
+
+			Dprintk("Setting variable MTRR %d, "
+				"base: %ld%cB, range: %ld%cB, type %s\n",
+				reg, start_base, start_factor,
+				size_base, size_factor,
+				(type == MTRR_TYPE_UNCACHABLE) ? "UC" :
+				   ((type == MTRR_TYPE_WRBACK) ? "WB" : "Other")
+				);
+		}
+		save_var_mtrr(reg++, range_startk, sizek, type);
+		range_startk += sizek;
+		range_sizek -= sizek;
+		if (reg >= num_var_ranges)
+			break;
+	}
+	return reg;
+}
+
+static unsigned __init
+range_to_mtrr_with_hole(struct var_mtrr_state *state, unsigned long basek,
+			unsigned long sizek)
+{
+	unsigned long hole_basek, hole_sizek;
+	unsigned long second_sizek;
+	unsigned long range0_basek, range0_sizek;
+	unsigned long range_basek, range_sizek;
+	unsigned long chunk_sizek;
+	unsigned long gran_sizek;
+
+	hole_basek = 0;
+	hole_sizek = 0;
+	second_sizek = 0;
+	chunk_sizek = state->chunk_sizek;
+	gran_sizek = state->gran_sizek;
+
+	/* Align with gran size, prevent small block used up MTRRs: */
+	range_basek = ALIGN(state->range_startk, gran_sizek);
+	if ((range_basek > basek) && basek)
+		return second_sizek;
+
+	state->range_sizek -= (range_basek - state->range_startk);
+	range_sizek = ALIGN(state->range_sizek, gran_sizek);
+
+	while (range_sizek > state->range_sizek) {
+		range_sizek -= gran_sizek;
+		if (!range_sizek)
+			return 0;
+	}
+	state->range_sizek = range_sizek;
+
+	/* Try to append some small hole: */
+	range0_basek = state->range_startk;
+	range0_sizek = ALIGN(state->range_sizek, chunk_sizek);
+
+	/* No increase: */
+	if (range0_sizek == state->range_sizek) {
+		Dprintk("rangeX: %016lx - %016lx\n",
+			range0_basek<<10,
+			(range0_basek + state->range_sizek)<<10);
+		state->reg = range_to_mtrr(state->reg, range0_basek,
+				state->range_sizek, MTRR_TYPE_WRBACK);
+		return 0;
+	}
+
+	/* Only cut back when it is not the last: */
+	if (sizek) {
+		while (range0_basek + range0_sizek > (basek + sizek)) {
+			if (range0_sizek >= chunk_sizek)
+				range0_sizek -= chunk_sizek;
+			else
+				range0_sizek = 0;
+
+			if (!range0_sizek)
+				break;
+		}
+	}
+
+second_try:
+	range_basek = range0_basek + range0_sizek;
+
+	/* One hole in the middle: */
+	if (range_basek > basek && range_basek <= (basek + sizek))
+		second_sizek = range_basek - basek;
+
+	if (range0_sizek > state->range_sizek) {
+
+		/* One hole in middle or at the end: */
+		hole_sizek = range0_sizek - state->range_sizek - second_sizek;
+
+		/* Hole size should be less than half of range0 size: */
+		if (hole_sizek >= (range0_sizek >> 1) &&
+		    range0_sizek >= chunk_sizek) {
+			range0_sizek -= chunk_sizek;
+			second_sizek = 0;
+			hole_sizek = 0;
+
+			goto second_try;
+		}
+	}
+
+	if (range0_sizek) {
+		Dprintk("range0: %016lx - %016lx\n",
+			range0_basek<<10,
+			(range0_basek + range0_sizek)<<10);
+		state->reg = range_to_mtrr(state->reg, range0_basek,
+				range0_sizek, MTRR_TYPE_WRBACK);
+	}
+
+	if (range0_sizek < state->range_sizek) {
+		/* Need to handle left over range: */
+		range_sizek = state->range_sizek - range0_sizek;
+
+		Dprintk("range: %016lx - %016lx\n",
+			 range_basek<<10,
+			 (range_basek + range_sizek)<<10);
+
+		state->reg = range_to_mtrr(state->reg, range_basek,
+				 range_sizek, MTRR_TYPE_WRBACK);
+	}
+
+	if (hole_sizek) {
+		hole_basek = range_basek - hole_sizek - second_sizek;
+		Dprintk("hole: %016lx - %016lx\n",
+			 hole_basek<<10,
+			 (hole_basek + hole_sizek)<<10);
+		state->reg = range_to_mtrr(state->reg, hole_basek,
+				 hole_sizek, MTRR_TYPE_UNCACHABLE);
+	}
+
+	return second_sizek;
+}
+
+static void __init
+set_var_mtrr_range(struct var_mtrr_state *state, unsigned long base_pfn,
+		   unsigned long size_pfn)
+{
+	unsigned long basek, sizek;
+	unsigned long second_sizek = 0;
+
+	if (state->reg >= num_var_ranges)
+		return;
+
+	basek = base_pfn << (PAGE_SHIFT - 10);
+	sizek = size_pfn << (PAGE_SHIFT - 10);
+
+	/* See if I can merge with the last range: */
+	if ((basek <= 1024) ||
+	    (state->range_startk + state->range_sizek == basek)) {
+		unsigned long endk = basek + sizek;
+		state->range_sizek = endk - state->range_startk;
+		return;
+	}
+	/* Write the range mtrrs: */
+	if (state->range_sizek != 0)
+		second_sizek = range_to_mtrr_with_hole(state, basek, sizek);
+
+	/* Allocate an msr: */
+	state->range_startk = basek + second_sizek;
+	state->range_sizek  = sizek - second_sizek;
+}
+
+/* Minimum size of mtrr block that can take hole: */
+static u64 mtrr_chunk_size __initdata = (256ULL<<20);
+
+static int __init parse_mtrr_chunk_size_opt(char *p)
+{
+	if (!p)
+		return -EINVAL;
+	mtrr_chunk_size = memparse(p, &p);
+	return 0;
+}
+early_param("mtrr_chunk_size", parse_mtrr_chunk_size_opt);
+
+/* Granularity of mtrr of block: */
+static u64 mtrr_gran_size __initdata;
+
+static int __init parse_mtrr_gran_size_opt(char *p)
+{
+	if (!p)
+		return -EINVAL;
+	mtrr_gran_size = memparse(p, &p);
+	return 0;
+}
+early_param("mtrr_gran_size", parse_mtrr_gran_size_opt);
+
+static unsigned long nr_mtrr_spare_reg __initdata =
+				 CONFIG_MTRR_SANITIZER_SPARE_REG_NR_DEFAULT;
+
+static int __init parse_mtrr_spare_reg(char *arg)
+{
+	if (!arg)
+		return -EINVAL;
+
+	return kstrtoul(arg, 0, &nr_mtrr_spare_reg);
+}
+early_param("mtrr_spare_reg_nr", parse_mtrr_spare_reg);
+
+static int __init
+x86_setup_var_mtrrs(struct range *range, int nr_range,
+		    u64 chunk_size, u64 gran_size)
+{
+	struct var_mtrr_state var_state;
+	int num_reg;
+	int i;
+
+	var_state.range_startk	= 0;
+	var_state.range_sizek	= 0;
+	var_state.reg		= 0;
+	var_state.chunk_sizek	= chunk_size >> 10;
+	var_state.gran_sizek	= gran_size >> 10;
+
+	memset(range_state, 0, sizeof(range_state));
+
+	/* Write the range: */
+	for (i = 0; i < nr_range; i++) {
+		set_var_mtrr_range(&var_state, range[i].start,
+				   range[i].end - range[i].start);
+	}
+
+	/* Write the last range: */
+	if (var_state.range_sizek != 0)
+		range_to_mtrr_with_hole(&var_state, 0, 0);
+
+	num_reg = var_state.reg;
+	/* Clear out the extra MTRR's: */
+	while (var_state.reg < num_var_ranges) {
+		save_var_mtrr(var_state.reg, 0, 0, 0);
+		var_state.reg++;
+	}
+
+	return num_reg;
+}
+
+struct mtrr_cleanup_result {
+	unsigned long	gran_sizek;
+	unsigned long	chunk_sizek;
+	unsigned long	lose_cover_sizek;
+	unsigned int	num_reg;
+	int		bad;
+};
+
+/*
+ * gran_size: 64K, 128K, 256K, 512K, 1M, 2M, ..., 2G
+ * chunk size: gran_size, ..., 2G
+ * so we need (1+16)*8
+ */
+#define NUM_RESULT	136
+#define PSHIFT		(PAGE_SHIFT - 10)
+
+static struct mtrr_cleanup_result __initdata result[NUM_RESULT];
+static unsigned long __initdata min_loss_pfn[RANGE_NUM];
+
+static void __init print_out_mtrr_range_state(void)
+{
+	char start_factor = 'K', size_factor = 'K';
+	unsigned long start_base, size_base;
+	mtrr_type type;
+	int i;
+
+	for (i = 0; i < num_var_ranges; i++) {
+
+		size_base = range_state[i].size_pfn << (PAGE_SHIFT - 10);
+		if (!size_base)
+			continue;
+
+		size_base = to_size_factor(size_base, &size_factor);
+		start_base = range_state[i].base_pfn << (PAGE_SHIFT - 10);
+		start_base = to_size_factor(start_base, &start_factor);
+		type = range_state[i].type;
+
+		Dprintk("reg %d, base: %ld%cB, range: %ld%cB, type %s\n",
+			i, start_base, start_factor,
+			size_base, size_factor,
+			(type == MTRR_TYPE_UNCACHABLE) ? "UC" :
+			    ((type == MTRR_TYPE_WRPROT) ? "WP" :
+			     ((type == MTRR_TYPE_WRBACK) ? "WB" : "Other"))
+			);
+	}
+}
+
+static int __init mtrr_need_cleanup(void)
+{
+	int i;
+	mtrr_type type;
+	unsigned long size;
+	/* Extra one for all 0: */
+	int num[MTRR_NUM_TYPES + 1];
+
+	/* Check entries number: */
+	memset(num, 0, sizeof(num));
+	for (i = 0; i < num_var_ranges; i++) {
+		type = range_state[i].type;
+		size = range_state[i].size_pfn;
+		if (type >= MTRR_NUM_TYPES)
+			continue;
+		if (!size)
+			type = MTRR_NUM_TYPES;
+		num[type]++;
+	}
+
+	/* Check if we got UC entries: */
+	if (!num[MTRR_TYPE_UNCACHABLE])
+		return 0;
+
+	/* Check if we only had WB and UC */
+	if (num[MTRR_TYPE_WRBACK] + num[MTRR_TYPE_UNCACHABLE] !=
+	    num_var_ranges - num[MTRR_NUM_TYPES])
+		return 0;
+
+	return 1;
+}
+
+static unsigned long __initdata range_sums;
+
+static void __init
+mtrr_calc_range_state(u64 chunk_size, u64 gran_size,
+		      unsigned long x_remove_base,
+		      unsigned long x_remove_size, int i)
+{
+	/*
+	 * range_new should really be an automatic variable, but
+	 * putting 4096 bytes on the stack is frowned upon, to put it
+	 * mildly. It is safe to make it a static __initdata variable,
+	 * since mtrr_calc_range_state is only called during init and
+	 * there's no way it will call itself recursively.
+	 */
+	static struct range range_new[RANGE_NUM] __initdata;
+	unsigned long range_sums_new;
+	int nr_range_new;
+	int num_reg;
+
+	/* Convert ranges to var ranges state: */
+	num_reg = x86_setup_var_mtrrs(range, nr_range, chunk_size, gran_size);
+
+	/* We got new setting in range_state, check it: */
+	memset(range_new, 0, sizeof(range_new));
+	nr_range_new = x86_get_mtrr_mem_range(range_new, 0,
+				x_remove_base, x_remove_size);
+	range_sums_new = sum_ranges(range_new, nr_range_new);
+
+	result[i].chunk_sizek = chunk_size >> 10;
+	result[i].gran_sizek = gran_size >> 10;
+	result[i].num_reg = num_reg;
+
+	if (range_sums < range_sums_new) {
+		result[i].lose_cover_sizek = (range_sums_new - range_sums) << PSHIFT;
+		result[i].bad = 1;
+	} else {
+		result[i].lose_cover_sizek = (range_sums - range_sums_new) << PSHIFT;
+	}
+
+	/* Double check it: */
+	if (!result[i].bad && !result[i].lose_cover_sizek) {
+		if (nr_range_new != nr_range || memcmp(range, range_new, sizeof(range)))
+			result[i].bad = 1;
+	}
+
+	if (!result[i].bad && (range_sums - range_sums_new < min_loss_pfn[num_reg]))
+		min_loss_pfn[num_reg] = range_sums - range_sums_new;
+}
+
+static void __init mtrr_print_out_one_result(int i)
+{
+	unsigned long gran_base, chunk_base, lose_base;
+	char gran_factor, chunk_factor, lose_factor;
+
+	gran_base = to_size_factor(result[i].gran_sizek, &gran_factor);
+	chunk_base = to_size_factor(result[i].chunk_sizek, &chunk_factor);
+	lose_base = to_size_factor(result[i].lose_cover_sizek, &lose_factor);
+
+	pr_info("%sgran_size: %ld%c \tchunk_size: %ld%c \t",
+		result[i].bad ? "*BAD*" : " ",
+		gran_base, gran_factor, chunk_base, chunk_factor);
+	pr_cont("num_reg: %d  \tlose cover RAM: %s%ld%c\n",
+		result[i].num_reg, result[i].bad ? "-" : "",
+		lose_base, lose_factor);
+}
+
+static int __init mtrr_search_optimal_index(void)
+{
+	int num_reg_good;
+	int index_good;
+	int i;
+
+	if (nr_mtrr_spare_reg >= num_var_ranges)
+		nr_mtrr_spare_reg = num_var_ranges - 1;
+
+	num_reg_good = -1;
+	for (i = num_var_ranges - nr_mtrr_spare_reg; i > 0; i--) {
+		if (!min_loss_pfn[i])
+			num_reg_good = i;
+	}
+
+	index_good = -1;
+	if (num_reg_good != -1) {
+		for (i = 0; i < NUM_RESULT; i++) {
+			if (!result[i].bad &&
+			    result[i].num_reg == num_reg_good &&
+			    !result[i].lose_cover_sizek) {
+				index_good = i;
+				break;
+			}
+		}
+	}
+
+	return index_good;
+}
+
+int __init mtrr_cleanup(void)
+{
+	unsigned long x_remove_base, x_remove_size;
+	unsigned long base, size, def, dummy;
+	u64 chunk_size, gran_size;
+	mtrr_type type;
+	int index_good;
+	int i;
+
+	if (!mtrr_enabled())
+		return 0;
+
+	if (!cpu_feature_enabled(X86_FEATURE_MTRR) || enable_mtrr_cleanup < 1)
+		return 0;
+
+	rdmsr(MSR_MTRRdefType, def, dummy);
+	def &= 0xff;
+	if (def != MTRR_TYPE_UNCACHABLE)
+		return 0;
+
+	/* Get it and store it aside: */
+	memset(range_state, 0, sizeof(range_state));
+	for (i = 0; i < num_var_ranges; i++) {
+		mtrr_if->get(i, &base, &size, &type);
+		range_state[i].base_pfn = base;
+		range_state[i].size_pfn = size;
+		range_state[i].type = type;
+	}
+
+	/* Check if we need handle it and can handle it: */
+	if (!mtrr_need_cleanup())
+		return 0;
+
+	/* Print original var MTRRs at first, for debugging: */
+	Dprintk("original variable MTRRs\n");
+	print_out_mtrr_range_state();
+
+	memset(range, 0, sizeof(range));
+	x_remove_size = 0;
+	x_remove_base = 1 << (32 - PAGE_SHIFT);
+	if (mtrr_tom2)
+		x_remove_size = (mtrr_tom2 >> PAGE_SHIFT) - x_remove_base;
+
+	/*
+	 * [0, 1M) should always be covered by var mtrr with WB
+	 * and fixed mtrrs should take effect before var mtrr for it:
+	 */
+	nr_range = add_range_with_merge(range, RANGE_NUM, 0, 0,
+					1ULL<<(20 - PAGE_SHIFT));
+	/* add from var mtrr at last */
+	nr_range = x86_get_mtrr_mem_range(range, nr_range,
+					  x_remove_base, x_remove_size);
+
+	range_sums = sum_ranges(range, nr_range);
+	pr_info("total RAM covered: %ldM\n",
+	       range_sums >> (20 - PAGE_SHIFT));
+
+	if (mtrr_chunk_size && mtrr_gran_size) {
+		i = 0;
+		mtrr_calc_range_state(mtrr_chunk_size, mtrr_gran_size,
+				      x_remove_base, x_remove_size, i);
+
+		mtrr_print_out_one_result(i);
+
+		if (!result[i].bad) {
+			set_var_mtrr_all();
+			Dprintk("New variable MTRRs\n");
+			print_out_mtrr_range_state();
+			return 1;
+		}
+		pr_info("invalid mtrr_gran_size or mtrr_chunk_size, will find optimal one\n");
+	}
+
+	i = 0;
+	memset(min_loss_pfn, 0xff, sizeof(min_loss_pfn));
+	memset(result, 0, sizeof(result));
+	for (gran_size = (1ULL<<16); gran_size < (1ULL<<32); gran_size <<= 1) {
+
+		for (chunk_size = gran_size; chunk_size < (1ULL<<32);
+		     chunk_size <<= 1) {
+
+			if (i >= NUM_RESULT)
+				continue;
+
+			mtrr_calc_range_state(chunk_size, gran_size,
+				      x_remove_base, x_remove_size, i);
+			if (mtrr_debug) {
+				mtrr_print_out_one_result(i);
+				pr_info("\n");
+			}
+
+			i++;
+		}
+	}
+
+	/* Try to find the optimal index: */
+	index_good = mtrr_search_optimal_index();
+
+	if (index_good != -1) {
+		pr_info("Found optimal setting for mtrr clean up\n");
+		i = index_good;
+		mtrr_print_out_one_result(i);
+
+		/* Convert ranges to var ranges state: */
+		chunk_size = result[i].chunk_sizek;
+		chunk_size <<= 10;
+		gran_size = result[i].gran_sizek;
+		gran_size <<= 10;
+		x86_setup_var_mtrrs(range, nr_range, chunk_size, gran_size);
+		set_var_mtrr_all();
+		Dprintk("New variable MTRRs\n");
+		print_out_mtrr_range_state();
+		return 1;
+	} else {
+		/* print out all */
+		for (i = 0; i < NUM_RESULT; i++)
+			mtrr_print_out_one_result(i);
+	}
+
+	pr_info("mtrr_cleanup: can not find optimal value\n");
+	pr_info("please specify mtrr_gran_size/mtrr_chunk_size\n");
+
+	return 0;
+}
+#else
+int __init mtrr_cleanup(void)
+{
+	return 0;
+}
+#endif
+
+static int disable_mtrr_trim;
+
+static int __init disable_mtrr_trim_setup(char *str)
+{
+	disable_mtrr_trim = 1;
+	return 0;
+}
+early_param("disable_mtrr_trim", disable_mtrr_trim_setup);
+
+/*
+ * Newer AMD K8s and later CPUs have a special magic MSR way to force WB
+ * for memory >4GB. Check for that here.
+ * Note this won't check if the MTRRs < 4GB where the magic bit doesn't
+ * apply to are wrong, but so far we don't know of any such case in the wild.
+ */
+#define Tom2Enabled		(1U << 21)
+#define Tom2ForceMemTypeWB	(1U << 22)
+
+int __init amd_special_default_mtrr(void)
+{
+	u32 l, h;
+
+	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD &&
+	    boot_cpu_data.x86_vendor != X86_VENDOR_HYGON)
+		return 0;
+	if (boot_cpu_data.x86 < 0xf)
+		return 0;
+	/* In case some hypervisor doesn't pass SYSCFG through: */
+	if (rdmsr_safe(MSR_AMD64_SYSCFG, &l, &h) < 0)
+		return 0;
+	/*
+	 * Memory between 4GB and top of mem is forced WB by this magic bit.
+	 * Reserved before K8RevF, but should be zero there.
+	 */
+	if ((l & (Tom2Enabled | Tom2ForceMemTypeWB)) ==
+		 (Tom2Enabled | Tom2ForceMemTypeWB))
+		return 1;
+	return 0;
+}
+
+static u64 __init
+real_trim_memory(unsigned long start_pfn, unsigned long limit_pfn)
+{
+	u64 trim_start, trim_size;
+
+	trim_start = start_pfn;
+	trim_start <<= PAGE_SHIFT;
+
+	trim_size = limit_pfn;
+	trim_size <<= PAGE_SHIFT;
+	trim_size -= trim_start;
+
+	return e820__range_update(trim_start, trim_size, E820_TYPE_RAM, E820_TYPE_RESERVED);
+}
+
+/**
+ * mtrr_trim_uncached_memory - trim RAM not covered by MTRRs
+ * @end_pfn: ending page frame number
+ *
+ * Some buggy BIOSes don't setup the MTRRs properly for systems with certain
+ * memory configurations.  This routine checks that the highest MTRR matches
+ * the end of memory, to make sure the MTRRs having a write back type cover
+ * all of the memory the kernel is intending to use.  If not, it'll trim any
+ * memory off the end by adjusting end_pfn, removing it from the kernel's
+ * allocation pools, warning the user with an obnoxious message.
+ */
+int __init mtrr_trim_uncached_memory(unsigned long end_pfn)
+{
+	unsigned long i, base, size, highest_pfn = 0, def, dummy;
+	mtrr_type type;
+	u64 total_trim_size;
+	/* extra one for all 0 */
+	int num[MTRR_NUM_TYPES + 1];
+
+	if (!mtrr_enabled())
+		return 0;
+
+	/*
+	 * Make sure we only trim uncachable memory on machines that
+	 * support the Intel MTRR architecture:
+	 */
+	if (!cpu_feature_enabled(X86_FEATURE_MTRR) || disable_mtrr_trim)
+		return 0;
+
+	rdmsr(MSR_MTRRdefType, def, dummy);
+	def &= MTRR_DEF_TYPE_TYPE;
+	if (def != MTRR_TYPE_UNCACHABLE)
+		return 0;
+
+	/* Get it and store it aside: */
+	memset(range_state, 0, sizeof(range_state));
+	for (i = 0; i < num_var_ranges; i++) {
+		mtrr_if->get(i, &base, &size, &type);
+		range_state[i].base_pfn = base;
+		range_state[i].size_pfn = size;
+		range_state[i].type = type;
+	}
+
+	/* Find highest cached pfn: */
+	for (i = 0; i < num_var_ranges; i++) {
+		type = range_state[i].type;
+		if (type != MTRR_TYPE_WRBACK)
+			continue;
+		base = range_state[i].base_pfn;
+		size = range_state[i].size_pfn;
+		if (highest_pfn < base + size)
+			highest_pfn = base + size;
+	}
+
+	/* kvm/qemu doesn't have mtrr set right, don't trim them all: */
+	if (!highest_pfn) {
+		pr_info("CPU MTRRs all blank - virtualized system.\n");
+		return 0;
+	}
+
+	/* Check entries number: */
+	memset(num, 0, sizeof(num));
+	for (i = 0; i < num_var_ranges; i++) {
+		type = range_state[i].type;
+		if (type >= MTRR_NUM_TYPES)
+			continue;
+		size = range_state[i].size_pfn;
+		if (!size)
+			type = MTRR_NUM_TYPES;
+		num[type]++;
+	}
+
+	/* No entry for WB? */
+	if (!num[MTRR_TYPE_WRBACK])
+		return 0;
+
+	/* Check if we only had WB and UC: */
+	if (num[MTRR_TYPE_WRBACK] + num[MTRR_TYPE_UNCACHABLE] !=
+		num_var_ranges - num[MTRR_NUM_TYPES])
+		return 0;
+
+	memset(range, 0, sizeof(range));
+	nr_range = 0;
+	if (mtrr_tom2) {
+		range[nr_range].start = (1ULL<<(32 - PAGE_SHIFT));
+		range[nr_range].end = mtrr_tom2 >> PAGE_SHIFT;
+		if (highest_pfn < range[nr_range].end)
+			highest_pfn = range[nr_range].end;
+		nr_range++;
+	}
+	nr_range = x86_get_mtrr_mem_range(range, nr_range, 0, 0);
+
+	/* Check the head: */
+	total_trim_size = 0;
+	if (range[0].start)
+		total_trim_size += real_trim_memory(0, range[0].start);
+
+	/* Check the holes: */
+	for (i = 0; i < nr_range - 1; i++) {
+		if (range[i].end < range[i+1].start)
+			total_trim_size += real_trim_memory(range[i].end,
+							    range[i+1].start);
+	}
+
+	/* Check the top: */
+	i = nr_range - 1;
+	if (range[i].end < end_pfn)
+		total_trim_size += real_trim_memory(range[i].end,
+							 end_pfn);
+
+	if (total_trim_size) {
+		pr_warn("WARNING: BIOS bug: CPU MTRRs don't cover all of memory, losing %lluMB of RAM.\n",
+			total_trim_size >> 20);
+
+		if (!changed_by_mtrr_cleanup)
+			WARN_ON(1);
+
+		pr_info("update e820 for mtrr\n");
+		e820__update_table_print();
+
+		return 1;
+	}
+
+	return 0;
+}

linux-7.2.3/arch/x86/kernel/cpu/mtrr/cyrix.c

@@ -0,0 +1,244 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/mm.h>
+
+#include <asm/processor-cyrix.h>
+#include <asm/processor-flags.h>
+#include <asm/mtrr.h>
+#include <asm/msr.h>
+
+#include "mtrr.h"
+
+static void
+cyrix_get_arr(unsigned int reg, unsigned long *base,
+	      unsigned long *size, mtrr_type * type)
+{
+	unsigned char arr, ccr3, rcr, shift;
+	unsigned long flags;
+
+	arr = CX86_ARR_BASE + (reg << 1) + reg;	/* avoid multiplication by 3 */
+
+	local_irq_save(flags);
+
+	ccr3 = getCx86(CX86_CCR3);
+	setCx86(CX86_CCR3, (ccr3 & 0x0f) | 0x10);	/* enable MAPEN */
+	((unsigned char *)base)[3] = getCx86(arr);
+	((unsigned char *)base)[2] = getCx86(arr + 1);
+	((unsigned char *)base)[1] = getCx86(arr + 2);
+	rcr = getCx86(CX86_RCR_BASE + reg);
+	setCx86(CX86_CCR3, ccr3);			/* disable MAPEN */
+
+	local_irq_restore(flags);
+
+	shift = ((unsigned char *) base)[1] & 0x0f;
+	*base >>= PAGE_SHIFT;
+
+	/*
+	 * Power of two, at least 4K on ARR0-ARR6, 256K on ARR7
+	 * Note: shift==0xf means 4G, this is unsupported.
+	 */
+	if (shift)
+		*size = (reg < 7 ? 0x1UL : 0x40UL) << (shift - 1);
+	else
+		*size = 0;
+
+	/* Bit 0 is Cache Enable on ARR7, Cache Disable on ARR0-ARR6 */
+	if (reg < 7) {
+		switch (rcr) {
+		case 1:
+			*type = MTRR_TYPE_UNCACHABLE;
+			break;
+		case 8:
+			*type = MTRR_TYPE_WRBACK;
+			break;
+		case 9:
+			*type = MTRR_TYPE_WRCOMB;
+			break;
+		case 24:
+		default:
+			*type = MTRR_TYPE_WRTHROUGH;
+			break;
+		}
+	} else {
+		switch (rcr) {
+		case 0:
+			*type = MTRR_TYPE_UNCACHABLE;
+			break;
+		case 8:
+			*type = MTRR_TYPE_WRCOMB;
+			break;
+		case 9:
+			*type = MTRR_TYPE_WRBACK;
+			break;
+		case 25:
+		default:
+			*type = MTRR_TYPE_WRTHROUGH;
+			break;
+		}
+	}
+}
+
+/*
+ * cyrix_get_free_region - get a free ARR.
+ *
+ * @base: the starting (base) address of the region.
+ * @size: the size (in bytes) of the region.
+ *
+ * Returns: the index of the region on success, else -1 on error.
+*/
+static int
+cyrix_get_free_region(unsigned long base, unsigned long size, int replace_reg)
+{
+	unsigned long lbase, lsize;
+	mtrr_type ltype;
+	int i;
+
+	switch (replace_reg) {
+	case 7:
+		if (size < 0x40)
+			break;
+		fallthrough;
+	case 6:
+	case 5:
+	case 4:
+		return replace_reg;
+	case 3:
+	case 2:
+	case 1:
+	case 0:
+		return replace_reg;
+	}
+	/* If we are to set up a region >32M then look at ARR7 immediately */
+	if (size > 0x2000) {
+		cyrix_get_arr(7, &lbase, &lsize, &ltype);
+		if (lsize == 0)
+			return 7;
+		/* Else try ARR0-ARR6 first  */
+	} else {
+		for (i = 0; i < 7; i++) {
+			cyrix_get_arr(i, &lbase, &lsize, &ltype);
+			if (lsize == 0)
+				return i;
+		}
+		/*
+		 * ARR0-ARR6 isn't free
+		 * try ARR7 but its size must be at least 256K
+		 */
+		cyrix_get_arr(i, &lbase, &lsize, &ltype);
+		if ((lsize == 0) && (size >= 0x40))
+			return i;
+	}
+	return -ENOSPC;
+}
+
+static u32 cr4, ccr3;
+
+static void prepare_set(void)
+{
+	u32 cr0;
+
+	/*  Save value of CR4 and clear Page Global Enable (bit 7)  */
+	if (boot_cpu_has(X86_FEATURE_PGE)) {
+		cr4 = __read_cr4();
+		__write_cr4(cr4 & ~X86_CR4_PGE);
+	}
+
+	/*
+	 * Disable and flush caches.
+	 * Note that wbinvd flushes the TLBs as a side-effect
+	 */
+	cr0 = read_cr0() | X86_CR0_CD;
+	wbinvd();
+	write_cr0(cr0);
+	wbinvd();
+
+	/* Cyrix ARRs - everything else was excluded at the top */
+	ccr3 = getCx86(CX86_CCR3);
+
+	/* Cyrix ARRs - everything else was excluded at the top */
+	setCx86(CX86_CCR3, (ccr3 & 0x0f) | 0x10);
+}
+
+static void post_set(void)
+{
+	/* Flush caches and TLBs */
+	wbinvd();
+
+	/* Cyrix ARRs - everything else was excluded at the top */
+	setCx86(CX86_CCR3, ccr3);
+
+	/* Enable caches */
+	write_cr0(read_cr0() & ~X86_CR0_CD);
+
+	/* Restore value of CR4 */
+	if (boot_cpu_has(X86_FEATURE_PGE))
+		__write_cr4(cr4);
+}
+
+static void cyrix_set_arr(unsigned int reg, unsigned long base,
+			  unsigned long size, mtrr_type type)
+{
+	unsigned char arr, arr_type, arr_size;
+
+	arr = CX86_ARR_BASE + (reg << 1) + reg;	/* avoid multiplication by 3 */
+
+	/* count down from 32M (ARR0-ARR6) or from 2G (ARR7) */
+	if (reg >= 7)
+		size >>= 6;
+
+	size &= 0x7fff;		/* make sure arr_size <= 14 */
+	for (arr_size = 0; size; arr_size++, size >>= 1)
+		;
+
+	if (reg < 7) {
+		switch (type) {
+		case MTRR_TYPE_UNCACHABLE:
+			arr_type = 1;
+			break;
+		case MTRR_TYPE_WRCOMB:
+			arr_type = 9;
+			break;
+		case MTRR_TYPE_WRTHROUGH:
+			arr_type = 24;
+			break;
+		default:
+			arr_type = 8;
+			break;
+		}
+	} else {
+		switch (type) {
+		case MTRR_TYPE_UNCACHABLE:
+			arr_type = 0;
+			break;
+		case MTRR_TYPE_WRCOMB:
+			arr_type = 8;
+			break;
+		case MTRR_TYPE_WRTHROUGH:
+			arr_type = 25;
+			break;
+		default:
+			arr_type = 9;
+			break;
+		}
+	}
+
+	prepare_set();
+
+	base <<= PAGE_SHIFT;
+	setCx86(arr + 0,  ((unsigned char *)&base)[3]);
+	setCx86(arr + 1,  ((unsigned char *)&base)[2]);
+	setCx86(arr + 2, (((unsigned char *)&base)[1]) | arr_size);
+	setCx86(CX86_RCR_BASE + reg, arr_type);
+
+	post_set();
+}
+
+const struct mtrr_ops cyrix_mtrr_ops = {
+	.var_regs          = 8,
+	.set               = cyrix_set_arr,
+	.get               = cyrix_get_arr,
+	.get_free_region   = cyrix_get_free_region,
+	.validate_add_page = generic_validate_add_page,
+	.have_wrcomb       = positive_have_wrcomb,
+};

linux-7.2.3/arch/x86/kernel/cpu/mtrr/if.c

@@ -0,0 +1,423 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/capability.h>
+#include <linux/seq_file.h>
+#include <linux/uaccess.h>
+#include <linux/proc_fs.h>
+#include <linux/ctype.h>
+#include <linux/string.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+
+#define LINE_SIZE 80
+
+#include <asm/mtrr.h>
+
+#include "mtrr.h"
+
+#define FILE_FCOUNT(f) (((struct seq_file *)((f)->private_data))->private)
+
+static const char *const mtrr_strings[MTRR_NUM_TYPES] =
+{
+	"uncachable",		/* 0 */
+	"write-combining",	/* 1 */
+	"?",			/* 2 */
+	"?",			/* 3 */
+	"write-through",	/* 4 */
+	"write-protect",	/* 5 */
+	"write-back",		/* 6 */
+};
+
+const char *mtrr_attrib_to_str(int x)
+{
+	return (x <= 6) ? mtrr_strings[x] : "?";
+}
+
+#ifdef CONFIG_PROC_FS
+
+static int
+mtrr_file_add(unsigned long base, unsigned long size,
+	      unsigned int type, bool increment, struct file *file, int page)
+{
+	unsigned int *fcount = FILE_FCOUNT(file);
+	int reg, max;
+
+	max = num_var_ranges;
+	if (fcount == NULL) {
+		fcount = kcalloc(max, sizeof(*fcount), GFP_KERNEL);
+		if (!fcount)
+			return -ENOMEM;
+		FILE_FCOUNT(file) = fcount;
+	}
+	if (!page) {
+		if ((base & (PAGE_SIZE - 1)) || (size & (PAGE_SIZE - 1)))
+			return -EINVAL;
+		base >>= PAGE_SHIFT;
+		size >>= PAGE_SHIFT;
+	}
+	reg = mtrr_add_page(base, size, type, true);
+	if (reg >= 0)
+		++fcount[reg];
+	return reg;
+}
+
+static int
+mtrr_file_del(unsigned long base, unsigned long size,
+	      struct file *file, int page)
+{
+	unsigned int *fcount = FILE_FCOUNT(file);
+	int reg;
+
+	if (!page) {
+		if ((base & (PAGE_SIZE - 1)) || (size & (PAGE_SIZE - 1)))
+			return -EINVAL;
+		base >>= PAGE_SHIFT;
+		size >>= PAGE_SHIFT;
+	}
+	reg = mtrr_del_page(-1, base, size);
+	if (reg < 0)
+		return reg;
+	if (fcount == NULL)
+		return reg;
+	if (fcount[reg] < 1)
+		return -EINVAL;
+	--fcount[reg];
+	return reg;
+}
+
+/*
+ * seq_file can seek but we ignore it.
+ *
+ * Format of control line:
+ *    "base=%Lx size=%Lx type=%s" or "disable=%d"
+ */
+static ssize_t
+mtrr_write(struct file *file, const char __user *buf, size_t len, loff_t * ppos)
+{
+	int i, err;
+	unsigned long reg;
+	unsigned long long base, size;
+	char *ptr;
+	char line[LINE_SIZE];
+	int length;
+
+	memset(line, 0, LINE_SIZE);
+
+	len = min_t(size_t, len, LINE_SIZE - 1);
+	length = strncpy_from_user(line, buf, len);
+	if (length < 0)
+		return length;
+
+	ptr = line + length - 1;
+	if (length && *ptr == '\n')
+		*ptr = '\0';
+
+	if (!strncmp(line, "disable=", 8)) {
+		reg = simple_strtoul(line + 8, &ptr, 0);
+		err = mtrr_del_page(reg, 0, 0);
+		if (err < 0)
+			return err;
+		return len;
+	}
+
+	if (strncmp(line, "base=", 5))
+		return -EINVAL;
+
+	base = simple_strtoull(line + 5, &ptr, 0);
+	ptr = skip_spaces(ptr);
+
+	if (strncmp(ptr, "size=", 5))
+		return -EINVAL;
+
+	size = simple_strtoull(ptr + 5, &ptr, 0);
+	if ((base & 0xfff) || (size & 0xfff))
+		return -EINVAL;
+	ptr = skip_spaces(ptr);
+
+	if (strncmp(ptr, "type=", 5))
+		return -EINVAL;
+	ptr = skip_spaces(ptr + 5);
+
+	i = match_string(mtrr_strings, MTRR_NUM_TYPES, ptr);
+	if (i < 0)
+		return i;
+
+	base >>= PAGE_SHIFT;
+	size >>= PAGE_SHIFT;
+	err = mtrr_add_page((unsigned long)base, (unsigned long)size, i, true);
+	if (err < 0)
+		return err;
+	return len;
+}
+
+static long
+mtrr_ioctl(struct file *file, unsigned int cmd, unsigned long __arg)
+{
+	int err = 0;
+	mtrr_type type;
+	unsigned long base;
+	unsigned long size;
+	struct mtrr_sentry sentry;
+	struct mtrr_gentry gentry;
+	void __user *arg = (void __user *) __arg;
+
+	memset(&gentry, 0, sizeof(gentry));
+
+	switch (cmd) {
+	case MTRRIOC_ADD_ENTRY:
+	case MTRRIOC_SET_ENTRY:
+	case MTRRIOC_DEL_ENTRY:
+	case MTRRIOC_KILL_ENTRY:
+	case MTRRIOC_ADD_PAGE_ENTRY:
+	case MTRRIOC_SET_PAGE_ENTRY:
+	case MTRRIOC_DEL_PAGE_ENTRY:
+	case MTRRIOC_KILL_PAGE_ENTRY:
+		if (copy_from_user(&sentry, arg, sizeof(sentry)))
+			return -EFAULT;
+		break;
+	case MTRRIOC_GET_ENTRY:
+	case MTRRIOC_GET_PAGE_ENTRY:
+		if (copy_from_user(&gentry, arg, sizeof(gentry)))
+			return -EFAULT;
+		break;
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_ADD_ENTRY:
+	case MTRRIOC32_SET_ENTRY:
+	case MTRRIOC32_DEL_ENTRY:
+	case MTRRIOC32_KILL_ENTRY:
+	case MTRRIOC32_ADD_PAGE_ENTRY:
+	case MTRRIOC32_SET_PAGE_ENTRY:
+	case MTRRIOC32_DEL_PAGE_ENTRY:
+	case MTRRIOC32_KILL_PAGE_ENTRY: {
+		struct mtrr_sentry32 __user *s32;
+
+		s32 = (struct mtrr_sentry32 __user *)__arg;
+		err = get_user(sentry.base, &s32->base);
+		err |= get_user(sentry.size, &s32->size);
+		err |= get_user(sentry.type, &s32->type);
+		if (err)
+			return err;
+		break;
+	}
+	case MTRRIOC32_GET_ENTRY:
+	case MTRRIOC32_GET_PAGE_ENTRY: {
+		struct mtrr_gentry32 __user *g32;
+
+		g32 = (struct mtrr_gentry32 __user *)__arg;
+		err = get_user(gentry.regnum, &g32->regnum);
+		err |= get_user(gentry.base, &g32->base);
+		err |= get_user(gentry.size, &g32->size);
+		err |= get_user(gentry.type, &g32->type);
+		if (err)
+			return err;
+		break;
+	}
+#endif
+	}
+
+	switch (cmd) {
+	default:
+		return -ENOTTY;
+	case MTRRIOC_ADD_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_ADD_ENTRY:
+#endif
+		err =
+		    mtrr_file_add(sentry.base, sentry.size, sentry.type, true,
+				  file, 0);
+		break;
+	case MTRRIOC_SET_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_SET_ENTRY:
+#endif
+		err = mtrr_add(sentry.base, sentry.size, sentry.type, false);
+		break;
+	case MTRRIOC_DEL_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_DEL_ENTRY:
+#endif
+		err = mtrr_file_del(sentry.base, sentry.size, file, 0);
+		break;
+	case MTRRIOC_KILL_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_KILL_ENTRY:
+#endif
+		err = mtrr_del(-1, sentry.base, sentry.size);
+		break;
+	case MTRRIOC_GET_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_GET_ENTRY:
+#endif
+		if (gentry.regnum >= num_var_ranges)
+			return -EINVAL;
+		mtrr_if->get(gentry.regnum, &base, &size, &type);
+
+		/* Hide entries that go above 4GB */
+		if (base + size - 1 >= (1UL << (8 * sizeof(gentry.size) - PAGE_SHIFT))
+		    || size >= (1UL << (8 * sizeof(gentry.size) - PAGE_SHIFT)))
+			gentry.base = gentry.size = gentry.type = 0;
+		else {
+			gentry.base = base << PAGE_SHIFT;
+			gentry.size = size << PAGE_SHIFT;
+			gentry.type = type;
+		}
+
+		break;
+	case MTRRIOC_ADD_PAGE_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_ADD_PAGE_ENTRY:
+#endif
+		err =
+		    mtrr_file_add(sentry.base, sentry.size, sentry.type, true,
+				  file, 1);
+		break;
+	case MTRRIOC_SET_PAGE_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_SET_PAGE_ENTRY:
+#endif
+		err =
+		    mtrr_add_page(sentry.base, sentry.size, sentry.type, false);
+		break;
+	case MTRRIOC_DEL_PAGE_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_DEL_PAGE_ENTRY:
+#endif
+		err = mtrr_file_del(sentry.base, sentry.size, file, 1);
+		break;
+	case MTRRIOC_KILL_PAGE_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_KILL_PAGE_ENTRY:
+#endif
+		err = mtrr_del_page(-1, sentry.base, sentry.size);
+		break;
+	case MTRRIOC_GET_PAGE_ENTRY:
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_GET_PAGE_ENTRY:
+#endif
+		if (gentry.regnum >= num_var_ranges)
+			return -EINVAL;
+		mtrr_if->get(gentry.regnum, &base, &size, &type);
+		/* Hide entries that would overflow */
+		if (size != (__typeof__(gentry.size))size)
+			gentry.base = gentry.size = gentry.type = 0;
+		else {
+			gentry.base = base;
+			gentry.size = size;
+			gentry.type = type;
+		}
+		break;
+	}
+
+	if (err)
+		return err;
+
+	switch (cmd) {
+	case MTRRIOC_GET_ENTRY:
+	case MTRRIOC_GET_PAGE_ENTRY:
+		if (copy_to_user(arg, &gentry, sizeof(gentry)))
+			err = -EFAULT;
+		break;
+#ifdef CONFIG_COMPAT
+	case MTRRIOC32_GET_ENTRY:
+	case MTRRIOC32_GET_PAGE_ENTRY: {
+		struct mtrr_gentry32 __user *g32;
+
+		g32 = (struct mtrr_gentry32 __user *)__arg;
+		err = put_user(gentry.base, &g32->base);
+		err |= put_user(gentry.size, &g32->size);
+		err |= put_user(gentry.regnum, &g32->regnum);
+		err |= put_user(gentry.type, &g32->type);
+		break;
+	}
+#endif
+	}
+	return err;
+}
+
+static int mtrr_close(struct inode *ino, struct file *file)
+{
+	unsigned int *fcount = FILE_FCOUNT(file);
+	int i, max;
+
+	if (fcount != NULL) {
+		max = num_var_ranges;
+		for (i = 0; i < max; ++i) {
+			while (fcount[i] > 0) {
+				mtrr_del(i, 0, 0);
+				--fcount[i];
+			}
+		}
+		kfree(fcount);
+		FILE_FCOUNT(file) = NULL;
+	}
+	return single_release(ino, file);
+}
+
+static int mtrr_seq_show(struct seq_file *seq, void *offset)
+{
+	char factor;
+	int i, max;
+	mtrr_type type;
+	unsigned long base, size;
+
+	max = num_var_ranges;
+	for (i = 0; i < max; i++) {
+		mtrr_if->get(i, &base, &size, &type);
+		if (size == 0) {
+			mtrr_usage_table[i] = 0;
+			continue;
+		}
+		if (size < (0x100000 >> PAGE_SHIFT)) {
+			/* less than 1MB */
+			factor = 'K';
+			size <<= PAGE_SHIFT - 10;
+		} else {
+			factor = 'M';
+			size >>= 20 - PAGE_SHIFT;
+		}
+		/* Base can be > 32bit */
+		seq_printf(seq, "reg%02i: base=0x%06lx000 (%5luMB), size=%5lu%cB, count=%d: %s\n",
+			   i, base, base >> (20 - PAGE_SHIFT),
+			   size, factor,
+			   mtrr_usage_table[i], mtrr_attrib_to_str(type));
+	}
+	return 0;
+}
+
+static int mtrr_open(struct inode *inode, struct file *file)
+{
+	if (!mtrr_if)
+		return -EIO;
+	if (!mtrr_if->get)
+		return -ENXIO;
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+	return single_open(file, mtrr_seq_show, NULL);
+}
+
+static const struct proc_ops mtrr_proc_ops = {
+	.proc_open		= mtrr_open,
+	.proc_read		= seq_read,
+	.proc_lseek		= seq_lseek,
+	.proc_write		= mtrr_write,
+	.proc_ioctl		= mtrr_ioctl,
+#ifdef CONFIG_COMPAT
+	.proc_compat_ioctl	= mtrr_ioctl,
+#endif
+	.proc_release		= mtrr_close,
+};
+
+static int __init mtrr_if_init(void)
+{
+	struct cpuinfo_x86 *c = &boot_cpu_data;
+
+	if ((!cpu_has(c, X86_FEATURE_MTRR)) &&
+	    (!cpu_has(c, X86_FEATURE_K6_MTRR)) &&
+	    (!cpu_has(c, X86_FEATURE_CYRIX_ARR)) &&
+	    (!cpu_has(c, X86_FEATURE_CENTAUR_MCR)))
+		return -ENODEV;
+
+	proc_create("mtrr", S_IWUSR | S_IRUGO, NULL, &mtrr_proc_ops);
+	return 0;
+}
+arch_initcall(mtrr_if_init);
+#endif			/*  CONFIG_PROC_FS  */

linux-7.2.3/arch/x86/kernel/cpu/mtrr/legacy.c

@@ -0,0 +1,94 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/syscore_ops.h>
+#include <asm/cpufeature.h>
+#include <asm/mtrr.h>
+#include <asm/processor.h>
+#include "mtrr.h"
+
+void mtrr_set_if(void)
+{
+	switch (boot_cpu_data.x86_vendor) {
+	case X86_VENDOR_AMD:
+		/* Pre-Athlon (K6) AMD CPU MTRRs */
+		if (cpu_feature_enabled(X86_FEATURE_K6_MTRR))
+			mtrr_if = &amd_mtrr_ops;
+		break;
+	case X86_VENDOR_CENTAUR:
+		if (cpu_feature_enabled(X86_FEATURE_CENTAUR_MCR))
+			mtrr_if = &centaur_mtrr_ops;
+		break;
+	case X86_VENDOR_CYRIX:
+		if (cpu_feature_enabled(X86_FEATURE_CYRIX_ARR))
+			mtrr_if = &cyrix_mtrr_ops;
+		break;
+	default:
+		break;
+	}
+}
+
+/*
+ * The suspend/resume methods are only for CPUs without MTRR. CPUs using generic
+ * MTRR driver don't require this.
+ */
+struct mtrr_value {
+	mtrr_type	ltype;
+	unsigned long	lbase;
+	unsigned long	lsize;
+};
+
+static struct mtrr_value *mtrr_value;
+
+static int mtrr_save(void *data)
+{
+	int i;
+
+	if (!mtrr_value)
+		return -ENOMEM;
+
+	for (i = 0; i < num_var_ranges; i++) {
+		mtrr_if->get(i, &mtrr_value[i].lbase,
+				&mtrr_value[i].lsize,
+				&mtrr_value[i].ltype);
+	}
+	return 0;
+}
+
+static void mtrr_restore(void *data)
+{
+	int i;
+
+	for (i = 0; i < num_var_ranges; i++) {
+		if (mtrr_value[i].lsize) {
+			mtrr_if->set(i, mtrr_value[i].lbase,
+				     mtrr_value[i].lsize,
+				     mtrr_value[i].ltype);
+		}
+	}
+}
+
+static const struct syscore_ops mtrr_syscore_ops = {
+	.suspend	= mtrr_save,
+	.resume		= mtrr_restore,
+};
+
+static struct syscore mtrr_syscore = {
+	.ops = &mtrr_syscore_ops,
+};
+
+void mtrr_register_syscore(void)
+{
+	mtrr_value = kzalloc_objs(*mtrr_value, num_var_ranges);
+
+	/*
+	 * The CPU has no MTRR and seems to not support SMP. They have
+	 * specific drivers, we use a tricky method to support
+	 * suspend/resume for them.
+	 *
+	 * TBD: is there any system with such CPU which supports
+	 * suspend/resume? If no, we should remove the code.
+	 */
+	register_syscore(&mtrr_syscore);
+}

linux-7.2.3/arch/x86/kernel/cpu/mtrr/mtrr.c

@@ -0,0 +1,633 @@
+// SPDX-License-Identifier: LGPL-2.0+
+/*  Generic MTRR (Memory Type Range Register) driver.
+
+    Copyright (C) 1997-2000  Richard Gooch
+    Copyright (c) 2002	     Patrick Mochel
+
+    Richard Gooch may be reached by email at  rgooch@atnf.csiro.au
+    The postal address is:
+      Richard Gooch, c/o ATNF, P. O. Box 76, Epping, N.S.W., 2121, Australia.
+
+    Source: "Pentium Pro Family Developer's Manual, Volume 3:
+    Operating System Writer's Guide" (Intel document number 242692),
+    section 11.11.7
+
+    This was cleaned and made readable by Patrick Mochel <mochel@osdl.org>
+    on 6-7 March 2002.
+    Source: Intel Architecture Software Developers Manual, Volume 3:
+    System Programming Guide; Section 9.11. (1997 edition - PPro).
+*/
+
+#include <linux/types.h> /* FIXME: kvm_para.h needs this */
+
+#include <linux/stop_machine.h>
+#include <linux/kvm_para.h>
+#include <linux/uaccess.h>
+#include <linux/export.h>
+#include <linux/mutex.h>
+#include <linux/init.h>
+#include <linux/sort.h>
+#include <linux/cpu.h>
+#include <linux/pci.h>
+#include <linux/smp.h>
+#include <linux/syscore_ops.h>
+#include <linux/rcupdate.h>
+
+#include <asm/cacheinfo.h>
+#include <asm/cpufeature.h>
+#include <asm/e820/api.h>
+#include <asm/mtrr.h>
+#include <asm/msr.h>
+#include <asm/memtype.h>
+
+#include "mtrr.h"
+
+static_assert(X86_MEMTYPE_UC == MTRR_TYPE_UNCACHABLE);
+static_assert(X86_MEMTYPE_WC == MTRR_TYPE_WRCOMB);
+static_assert(X86_MEMTYPE_WT == MTRR_TYPE_WRTHROUGH);
+static_assert(X86_MEMTYPE_WP == MTRR_TYPE_WRPROT);
+static_assert(X86_MEMTYPE_WB == MTRR_TYPE_WRBACK);
+
+/* arch_phys_wc_add returns an MTRR register index plus this offset. */
+#define MTRR_TO_PHYS_WC_OFFSET 1000
+
+u32 num_var_ranges;
+
+unsigned int mtrr_usage_table[MTRR_MAX_VAR_RANGES];
+DEFINE_MUTEX(mtrr_mutex);
+
+const struct mtrr_ops *mtrr_if;
+
+/*  Returns non-zero if we have the write-combining memory type  */
+static int have_wrcomb(void)
+{
+	struct pci_dev *dev;
+
+	dev = pci_get_class(PCI_CLASS_BRIDGE_HOST << 8, NULL);
+	if (dev != NULL) {
+		/*
+		 * ServerWorks LE chipsets < rev 6 have problems with
+		 * write-combining. Don't allow it and leave room for other
+		 * chipsets to be tagged
+		 */
+		if (dev->vendor == PCI_VENDOR_ID_SERVERWORKS &&
+		    dev->device == PCI_DEVICE_ID_SERVERWORKS_LE &&
+		    dev->revision <= 5) {
+			pr_info("Serverworks LE rev < 6 detected. Write-combining disabled.\n");
+			pci_dev_put(dev);
+			return 0;
+		}
+		/*
+		 * Intel 450NX errata # 23. Non ascending cacheline evictions to
+		 * write combining memory may resulting in data corruption
+		 */
+		if (dev->vendor == PCI_VENDOR_ID_INTEL &&
+		    dev->device == PCI_DEVICE_ID_INTEL_82451NX) {
+			pr_info("Intel 450NX MMC detected. Write-combining disabled.\n");
+			pci_dev_put(dev);
+			return 0;
+		}
+		pci_dev_put(dev);
+	}
+	return mtrr_if->have_wrcomb ? mtrr_if->have_wrcomb() : 0;
+}
+
+static void __init init_table(void)
+{
+	int i, max;
+
+	max = num_var_ranges;
+	for (i = 0; i < max; i++)
+		mtrr_usage_table[i] = 1;
+}
+
+struct set_mtrr_data {
+	unsigned long	smp_base;
+	unsigned long	smp_size;
+	unsigned int	smp_reg;
+	mtrr_type	smp_type;
+};
+
+/**
+ * mtrr_rendezvous_handler - Work done in the synchronization handler. Executed
+ * by all the CPUs.
+ * @info: pointer to mtrr configuration data
+ *
+ * Returns nothing.
+ */
+static int mtrr_rendezvous_handler(void *info)
+{
+	struct set_mtrr_data *data = info;
+
+	mtrr_if->set(data->smp_reg, data->smp_base,
+		     data->smp_size, data->smp_type);
+	return 0;
+}
+
+static inline int types_compatible(mtrr_type type1, mtrr_type type2)
+{
+	return type1 == MTRR_TYPE_UNCACHABLE ||
+	       type2 == MTRR_TYPE_UNCACHABLE ||
+	       (type1 == MTRR_TYPE_WRTHROUGH && type2 == MTRR_TYPE_WRBACK) ||
+	       (type1 == MTRR_TYPE_WRBACK && type2 == MTRR_TYPE_WRTHROUGH);
+}
+
+/**
+ * set_mtrr - update mtrrs on all processors
+ * @reg:	mtrr in question
+ * @base:	mtrr base
+ * @size:	mtrr size
+ * @type:	mtrr type
+ *
+ * This is kinda tricky, but fortunately, Intel spelled it out for us cleanly:
+ *
+ * 1. Queue work to do the following on all processors:
+ * 2. Disable Interrupts
+ * 3. Wait for all procs to do so
+ * 4. Enter no-fill cache mode
+ * 5. Flush caches
+ * 6. Clear PGE bit
+ * 7. Flush all TLBs
+ * 8. Disable all range registers
+ * 9. Update the MTRRs
+ * 10. Enable all range registers
+ * 11. Flush all TLBs and caches again
+ * 12. Enter normal cache mode and reenable caching
+ * 13. Set PGE
+ * 14. Wait for buddies to catch up
+ * 15. Enable interrupts.
+ *
+ * What does that mean for us? Well, stop_machine() will ensure that
+ * the rendezvous handler is started on each CPU. And in lockstep they
+ * do the state transition of disabling interrupts, updating MTRR's
+ * (the CPU vendors may each do it differently, so we call mtrr_if->set()
+ * callback and let them take care of it.) and enabling interrupts.
+ *
+ * Note that the mechanism is the same for UP systems, too; all the SMP stuff
+ * becomes nops.
+ */
+static void set_mtrr(unsigned int reg, unsigned long base, unsigned long size,
+		     mtrr_type type)
+{
+	struct set_mtrr_data data = { .smp_reg = reg,
+				      .smp_base = base,
+				      .smp_size = size,
+				      .smp_type = type
+				    };
+
+	stop_machine_cpuslocked(mtrr_rendezvous_handler, &data, cpu_online_mask);
+
+	generic_rebuild_map();
+}
+
+/**
+ * mtrr_add_page - Add a memory type region
+ * @base: Physical base address of region in pages (in units of 4 kB!)
+ * @size: Physical size of region in pages (4 kB)
+ * @type: Type of MTRR desired
+ * @increment: If this is true do usage counting on the region
+ *
+ * Memory type region registers control the caching on newer Intel and
+ * non Intel processors. This function allows drivers to request an
+ * MTRR is added. The details and hardware specifics of each processor's
+ * implementation are hidden from the caller, but nevertheless the
+ * caller should expect to need to provide a power of two size on an
+ * equivalent power of two boundary.
+ *
+ * If the region cannot be added either because all regions are in use
+ * or the CPU cannot support it a negative value is returned. On success
+ * the register number for this entry is returned, but should be treated
+ * as a cookie only.
+ *
+ * On a multiprocessor machine the changes are made to all processors.
+ * This is required on x86 by the Intel processors.
+ *
+ * The available types are
+ *
+ * %MTRR_TYPE_UNCACHABLE - No caching
+ *
+ * %MTRR_TYPE_WRBACK - Write data back in bursts whenever
+ *
+ * %MTRR_TYPE_WRCOMB - Write data back soon but allow bursts
+ *
+ * %MTRR_TYPE_WRTHROUGH - Cache reads but not writes
+ *
+ * BUGS: Needs a quiet flag for the cases where drivers do not mind
+ * failures and do not wish system log messages to be sent.
+ */
+int mtrr_add_page(unsigned long base, unsigned long size,
+		  unsigned int type, bool increment)
+{
+	unsigned long lbase, lsize;
+	int i, replace, error;
+	mtrr_type ltype;
+
+	if (!mtrr_enabled())
+		return -ENXIO;
+
+	error = mtrr_if->validate_add_page(base, size, type);
+	if (error)
+		return error;
+
+	if (type >= MTRR_NUM_TYPES) {
+		pr_warn("type: %u invalid\n", type);
+		return -EINVAL;
+	}
+
+	/* If the type is WC, check that this processor supports it */
+	if ((type == MTRR_TYPE_WRCOMB) && !have_wrcomb()) {
+		pr_warn("your processor doesn't support write-combining\n");
+		return -ENOSYS;
+	}
+
+	if (!size) {
+		pr_warn("zero sized request\n");
+		return -EINVAL;
+	}
+
+	if ((base | (base + size - 1)) >>
+	    (boot_cpu_data.x86_phys_bits - PAGE_SHIFT)) {
+		pr_warn("base or size exceeds the MTRR width\n");
+		return -EINVAL;
+	}
+
+	error = -EINVAL;
+	replace = -1;
+
+	/* No CPU hotplug when we change MTRR entries */
+	cpus_read_lock();
+
+	/* Search for existing MTRR  */
+	mutex_lock(&mtrr_mutex);
+	for (i = 0; i < num_var_ranges; ++i) {
+		mtrr_if->get(i, &lbase, &lsize, &ltype);
+		if (!lsize || base > lbase + lsize - 1 ||
+		    base + size - 1 < lbase)
+			continue;
+		/*
+		 * At this point we know there is some kind of
+		 * overlap/enclosure
+		 */
+		if (base < lbase || base + size - 1 > lbase + lsize - 1) {
+			if (base <= lbase &&
+			    base + size - 1 >= lbase + lsize - 1) {
+				/*  New region encloses an existing region  */
+				if (type == ltype) {
+					replace = replace == -1 ? i : -2;
+					continue;
+				} else if (types_compatible(type, ltype))
+					continue;
+			}
+			pr_warn("0x%lx000,0x%lx000 overlaps existing 0x%lx000,0x%lx000\n", base, size, lbase,
+				lsize);
+			goto out;
+		}
+		/* New region is enclosed by an existing region */
+		if (ltype != type) {
+			if (types_compatible(type, ltype))
+				continue;
+			pr_warn("type mismatch for %lx000,%lx000 old: %s new: %s\n",
+				base, size, mtrr_attrib_to_str(ltype),
+				mtrr_attrib_to_str(type));
+			goto out;
+		}
+		if (increment)
+			++mtrr_usage_table[i];
+		error = i;
+		goto out;
+	}
+	/* Search for an empty MTRR */
+	i = mtrr_if->get_free_region(base, size, replace);
+	if (i >= 0) {
+		set_mtrr(i, base, size, type);
+		if (likely(replace < 0)) {
+			mtrr_usage_table[i] = 1;
+		} else {
+			mtrr_usage_table[i] = mtrr_usage_table[replace];
+			if (increment)
+				mtrr_usage_table[i]++;
+			if (unlikely(replace != i)) {
+				set_mtrr(replace, 0, 0, 0);
+				mtrr_usage_table[replace] = 0;
+			}
+		}
+	} else {
+		pr_info("no more MTRRs available\n");
+	}
+	error = i;
+ out:
+	mutex_unlock(&mtrr_mutex);
+	cpus_read_unlock();
+	return error;
+}
+
+static int mtrr_check(unsigned long base, unsigned long size)
+{
+	if ((base & (PAGE_SIZE - 1)) || (size & (PAGE_SIZE - 1))) {
+		pr_warn("size and base must be multiples of 4 kiB\n");
+		Dprintk("size: 0x%lx  base: 0x%lx\n", size, base);
+		dump_stack();
+		return -1;
+	}
+	return 0;
+}
+
+/**
+ * mtrr_add - Add a memory type region
+ * @base: Physical base address of region
+ * @size: Physical size of region
+ * @type: Type of MTRR desired
+ * @increment: If this is true do usage counting on the region
+ *
+ * Memory type region registers control the caching on newer Intel and
+ * non Intel processors. This function allows drivers to request an
+ * MTRR is added. The details and hardware specifics of each processor's
+ * implementation are hidden from the caller, but nevertheless the
+ * caller should expect to need to provide a power of two size on an
+ * equivalent power of two boundary.
+ *
+ * If the region cannot be added either because all regions are in use
+ * or the CPU cannot support it a negative value is returned. On success
+ * the register number for this entry is returned, but should be treated
+ * as a cookie only.
+ *
+ * On a multiprocessor machine the changes are made to all processors.
+ * This is required on x86 by the Intel processors.
+ *
+ * The available types are
+ *
+ * %MTRR_TYPE_UNCACHABLE - No caching
+ *
+ * %MTRR_TYPE_WRBACK - Write data back in bursts whenever
+ *
+ * %MTRR_TYPE_WRCOMB - Write data back soon but allow bursts
+ *
+ * %MTRR_TYPE_WRTHROUGH - Cache reads but not writes
+ *
+ * BUGS: Needs a quiet flag for the cases where drivers do not mind
+ * failures and do not wish system log messages to be sent.
+ */
+int mtrr_add(unsigned long base, unsigned long size, unsigned int type,
+	     bool increment)
+{
+	if (!mtrr_enabled())
+		return -ENODEV;
+	if (mtrr_check(base, size))
+		return -EINVAL;
+	return mtrr_add_page(base >> PAGE_SHIFT, size >> PAGE_SHIFT, type,
+			     increment);
+}
+
+/**
+ * mtrr_del_page - delete a memory type region
+ * @reg: Register returned by mtrr_add
+ * @base: Physical base address
+ * @size: Size of region
+ *
+ * If register is supplied then base and size are ignored. This is
+ * how drivers should call it.
+ *
+ * Releases an MTRR region. If the usage count drops to zero the
+ * register is freed and the region returns to default state.
+ * On success the register is returned, on failure a negative error
+ * code.
+ */
+int mtrr_del_page(int reg, unsigned long base, unsigned long size)
+{
+	int i, max;
+	mtrr_type ltype;
+	unsigned long lbase, lsize;
+	int error = -EINVAL;
+
+	if (!mtrr_enabled())
+		return -ENODEV;
+
+	max = num_var_ranges;
+	/* No CPU hotplug when we change MTRR entries */
+	cpus_read_lock();
+	mutex_lock(&mtrr_mutex);
+	if (reg < 0) {
+		/*  Search for existing MTRR  */
+		for (i = 0; i < max; ++i) {
+			mtrr_if->get(i, &lbase, &lsize, &ltype);
+			if (lbase == base && lsize == size) {
+				reg = i;
+				break;
+			}
+		}
+		if (reg < 0) {
+			Dprintk("no MTRR for %lx000,%lx000 found\n", base, size);
+			goto out;
+		}
+	}
+	if (reg >= max) {
+		pr_warn("register: %d too big\n", reg);
+		goto out;
+	}
+	mtrr_if->get(reg, &lbase, &lsize, &ltype);
+	if (lsize < 1) {
+		pr_warn("MTRR %d not used\n", reg);
+		goto out;
+	}
+	if (mtrr_usage_table[reg] < 1) {
+		pr_warn("reg: %d has count=0\n", reg);
+		goto out;
+	}
+	if (--mtrr_usage_table[reg] < 1)
+		set_mtrr(reg, 0, 0, 0);
+	error = reg;
+ out:
+	mutex_unlock(&mtrr_mutex);
+	cpus_read_unlock();
+	return error;
+}
+
+/**
+ * mtrr_del - delete a memory type region
+ * @reg: Register returned by mtrr_add
+ * @base: Physical base address
+ * @size: Size of region
+ *
+ * If register is supplied then base and size are ignored. This is
+ * how drivers should call it.
+ *
+ * Releases an MTRR region. If the usage count drops to zero the
+ * register is freed and the region returns to default state.
+ * On success the register is returned, on failure a negative error
+ * code.
+ */
+int mtrr_del(int reg, unsigned long base, unsigned long size)
+{
+	if (!mtrr_enabled())
+		return -ENODEV;
+	if (mtrr_check(base, size))
+		return -EINVAL;
+	return mtrr_del_page(reg, base >> PAGE_SHIFT, size >> PAGE_SHIFT);
+}
+
+/**
+ * arch_phys_wc_add - add a WC MTRR and handle errors if PAT is unavailable
+ * @base: Physical base address
+ * @size: Size of region
+ *
+ * If PAT is available, this does nothing.  If PAT is unavailable, it
+ * attempts to add a WC MTRR covering size bytes starting at base and
+ * logs an error if this fails.
+ *
+ * The called should provide a power of two size on an equivalent
+ * power of two boundary.
+ *
+ * Drivers must store the return value to pass to mtrr_del_wc_if_needed,
+ * but drivers should not try to interpret that return value.
+ */
+int arch_phys_wc_add(unsigned long base, unsigned long size)
+{
+	int ret;
+
+	if (pat_enabled() || !mtrr_enabled())
+		return 0;  /* Success!  (We don't need to do anything.) */
+
+	ret = mtrr_add(base, size, MTRR_TYPE_WRCOMB, true);
+	if (ret < 0) {
+		pr_warn("Failed to add WC MTRR for [%p-%p]; performance may suffer.",
+			(void *)base, (void *)(base + size - 1));
+		return ret;
+	}
+	return ret + MTRR_TO_PHYS_WC_OFFSET;
+}
+EXPORT_SYMBOL(arch_phys_wc_add);
+
+/*
+ * arch_phys_wc_del - undoes arch_phys_wc_add
+ * @handle: Return value from arch_phys_wc_add
+ *
+ * This cleans up after mtrr_add_wc_if_needed.
+ *
+ * The API guarantees that mtrr_del_wc_if_needed(error code) and
+ * mtrr_del_wc_if_needed(0) do nothing.
+ */
+void arch_phys_wc_del(int handle)
+{
+	if (handle >= 1) {
+		WARN_ON(handle < MTRR_TO_PHYS_WC_OFFSET);
+		mtrr_del(handle - MTRR_TO_PHYS_WC_OFFSET, 0, 0);
+	}
+}
+EXPORT_SYMBOL(arch_phys_wc_del);
+
+/*
+ * arch_phys_wc_index - translates arch_phys_wc_add's return value
+ * @handle: Return value from arch_phys_wc_add
+ *
+ * This will turn the return value from arch_phys_wc_add into an mtrr
+ * index suitable for debugging.
+ *
+ * Note: There is no legitimate use for this function, except possibly
+ * in printk line.  Alas there is an illegitimate use in some ancient
+ * drm ioctls.
+ */
+int arch_phys_wc_index(int handle)
+{
+	if (handle < MTRR_TO_PHYS_WC_OFFSET)
+		return -1;
+	else
+		return handle - MTRR_TO_PHYS_WC_OFFSET;
+}
+EXPORT_SYMBOL_GPL(arch_phys_wc_index);
+
+int __initdata changed_by_mtrr_cleanup;
+
+/**
+ * mtrr_bp_init - initialize MTRRs on the boot CPU
+ *
+ * This needs to be called early; before any of the other CPUs are
+ * initialized (i.e. before smp_init()).
+ */
+void __init mtrr_bp_init(void)
+{
+	bool generic_mtrrs = cpu_feature_enabled(X86_FEATURE_MTRR);
+	const char *why = "(not available)";
+	unsigned long config, dummy;
+
+	phys_hi_rsvd = GENMASK(31, boot_cpu_data.x86_phys_bits - 32);
+
+	if (!generic_mtrrs && mtrr_state.enabled) {
+		/*
+		 * Software overwrite of MTRR state, only for generic case.
+		 * Note that X86_FEATURE_MTRR has been reset in this case.
+		 */
+		init_table();
+		mtrr_build_map();
+		pr_info("MTRRs set to read-only\n");
+
+		return;
+	}
+
+	if (generic_mtrrs)
+		mtrr_if = &generic_mtrr_ops;
+	else
+		mtrr_set_if();
+
+	if (mtrr_enabled()) {
+		/* Get the number of variable MTRR ranges. */
+		if (mtrr_if == &generic_mtrr_ops)
+			rdmsr(MSR_MTRRcap, config, dummy);
+		else
+			config = mtrr_if->var_regs;
+		num_var_ranges = config & MTRR_CAP_VCNT;
+
+		init_table();
+		if (mtrr_if == &generic_mtrr_ops) {
+			/* BIOS may override */
+			if (get_mtrr_state()) {
+				memory_caching_control |= CACHE_MTRR;
+				changed_by_mtrr_cleanup = mtrr_cleanup();
+				mtrr_build_map();
+			} else {
+				mtrr_if = NULL;
+				why = "by BIOS";
+			}
+		}
+	}
+
+	if (!mtrr_enabled())
+		pr_info("MTRRs disabled %s\n", why);
+}
+
+/**
+ * mtrr_save_state - Save current fixed-range MTRR state of the first
+ *	cpu in cpu_online_mask.
+ */
+void mtrr_save_state(void)
+{
+	int first_cpu;
+
+	if (!mtrr_enabled() || !mtrr_state.have_fixed)
+		return;
+
+	first_cpu = cpumask_first(cpu_online_mask);
+	smp_call_function_single(first_cpu, mtrr_save_fixed_ranges, NULL, 1);
+}
+
+static int __init mtrr_init_finalize(void)
+{
+	/*
+	 * Map might exist if guest_force_mtrr_state() has been called or if
+	 * mtrr_enabled() returns true.
+	 */
+	mtrr_copy_map();
+
+	if (!mtrr_enabled())
+		return 0;
+
+	if (memory_caching_control & CACHE_MTRR) {
+		if (!changed_by_mtrr_cleanup)
+			mtrr_state_warn();
+		return 0;
+	}
+
+	mtrr_register_syscore();
+
+	return 0;
+}
+subsys_initcall(mtrr_init_finalize);

linux-7.2.3/arch/x86/kernel/cpu/mtrr/mtrr.h

@@ -0,0 +1,90 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * local MTRR defines.
+ */
+
+#include <linux/types.h>
+#include <linux/stddef.h>
+
+#define MTRR_CHANGE_MASK_FIXED     0x01
+#define MTRR_CHANGE_MASK_VARIABLE  0x02
+#define MTRR_CHANGE_MASK_DEFTYPE   0x04
+
+extern bool mtrr_debug;
+#define Dprintk(x...) do { if (mtrr_debug) pr_info(x); } while (0)
+
+extern unsigned int mtrr_usage_table[MTRR_MAX_VAR_RANGES];
+
+struct mtrr_ops {
+	u32	var_regs;
+	void	(*set)(unsigned int reg, unsigned long base,
+		       unsigned long size, mtrr_type type);
+	void	(*get)(unsigned int reg, unsigned long *base,
+		       unsigned long *size, mtrr_type *type);
+	int	(*get_free_region)(unsigned long base, unsigned long size,
+				   int replace_reg);
+	int	(*validate_add_page)(unsigned long base, unsigned long size,
+				     unsigned int type);
+	int	(*have_wrcomb)(void);
+};
+
+extern int generic_get_free_region(unsigned long base, unsigned long size,
+				   int replace_reg);
+extern int generic_validate_add_page(unsigned long base, unsigned long size,
+				     unsigned int type);
+
+extern const struct mtrr_ops generic_mtrr_ops;
+
+extern int positive_have_wrcomb(void);
+
+/* library functions for processor-specific routines */
+struct set_mtrr_context {
+	unsigned long	flags;
+	unsigned long	cr4val;
+	u32		deftype_lo;
+	u32		deftype_hi;
+	u32		ccr3;
+};
+
+void fill_mtrr_var_range(unsigned int index,
+		u32 base_lo, u32 base_hi, u32 mask_lo, u32 mask_hi);
+bool get_mtrr_state(void);
+
+extern const struct mtrr_ops *mtrr_if;
+extern struct mutex mtrr_mutex;
+
+extern unsigned int num_var_ranges;
+extern u64 mtrr_tom2;
+extern struct mtrr_state_type mtrr_state;
+extern u32 phys_hi_rsvd;
+
+void mtrr_state_warn(void);
+const char *mtrr_attrib_to_str(int x);
+void mtrr_wrmsr(unsigned, unsigned, unsigned);
+#ifdef CONFIG_X86_32
+void mtrr_set_if(void);
+void mtrr_register_syscore(void);
+#else
+static inline void mtrr_set_if(void) { }
+static inline void mtrr_register_syscore(void) { }
+#endif
+void mtrr_build_map(void);
+void mtrr_copy_map(void);
+
+/* CPU specific mtrr_ops vectors. */
+extern const struct mtrr_ops amd_mtrr_ops;
+extern const struct mtrr_ops cyrix_mtrr_ops;
+extern const struct mtrr_ops centaur_mtrr_ops;
+
+extern int changed_by_mtrr_cleanup;
+extern int mtrr_cleanup(void);
+
+/*
+ * Must be used by code which uses mtrr_if to call platform-specific
+ * MTRR manipulation functions.
+ */
+static inline bool mtrr_enabled(void)
+{
+	return !!mtrr_if;
+}
+void generic_rebuild_map(void);