[deliverable/linux.git] / arch / x86 / kernel / crash.c

/*
 * Architecture specific (i386/x86_64) functions for kexec based crash dumps.
 *
 * Created by: Hariprasad Nellitheertha (hari@in.ibm.com)
 *
 * Copyright (C) IBM Corporation, 2004. All rights reserved.
 * Copyright (C) Red Hat Inc., 2014. All rights reserved.
 * Authors:
 *      Vivek Goyal <vgoyal@redhat.com>
 *
 */

#define pr_fmt(fmt)	"kexec: " fmt

#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/smp.h>
#include <linux/reboot.h>
#include <linux/kexec.h>
#include <linux/delay.h>
#include <linux/elf.h>
#include <linux/elfcore.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>

#include <asm/processor.h>
#include <asm/hardirq.h>
#include <asm/nmi.h>
#include <asm/hw_irq.h>
#include <asm/apic.h>
#include <asm/io_apic.h>
#include <asm/hpet.h>
#include <linux/kdebug.h>
#include <asm/cpu.h>
#include <asm/reboot.h>
#include <asm/virtext.h>
#include <asm/intel_pt.h>

/* Alignment required for elf header segment */
#define ELF_CORE_HEADER_ALIGN   4096

/* This primarily represents number of split ranges due to exclusion */
#define CRASH_MAX_RANGES	16

struct crash_mem_range {
	u64 start, end;
};

struct crash_mem {
	unsigned int nr_ranges;
	struct crash_mem_range ranges[CRASH_MAX_RANGES];
};

/* Misc data about ram ranges needed to prepare elf headers */
struct crash_elf_data {
	struct kimage *image;
	/*
	 * Total number of ram ranges we have after various adjustments for
	 * GART, crash reserved region etc.
	 */
	unsigned int max_nr_ranges;
	unsigned long gart_start, gart_end;

	/* Pointer to elf header */
	void *ehdr;
	/* Pointer to next phdr */
	void *bufp;
	struct crash_mem mem;
};

/* Used while preparing memory map entries for second kernel */
struct crash_memmap_data {
	struct boot_params *params;
	/* Type of memory */
	unsigned int type;
};

/*
 * This is used to VMCLEAR all VMCSs loaded on the
 * processor. And when loading kvm_intel module, the
 * callback function pointer will be assigned.
 *
 * protected by rcu.
 */
crash_vmclear_fn __rcu *crash_vmclear_loaded_vmcss = NULL;
EXPORT_SYMBOL_GPL(crash_vmclear_loaded_vmcss);
unsigned long crash_zero_bytes;

static inline void cpu_crash_vmclear_loaded_vmcss(void)
{
	crash_vmclear_fn *do_vmclear_operation = NULL;

	rcu_read_lock();
	do_vmclear_operation = rcu_dereference(crash_vmclear_loaded_vmcss);
	if (do_vmclear_operation)
		do_vmclear_operation();
	rcu_read_unlock();
}

#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC)

static void kdump_nmi_callback(int cpu, struct pt_regs *regs)
{
#ifdef CONFIG_X86_32
	struct pt_regs fixed_regs;

	if (!user_mode(regs)) {
		crash_fixup_ss_esp(&fixed_regs, regs);
		regs = &fixed_regs;
	}
#endif
	crash_save_cpu(regs, cpu);

	/*
	 * VMCLEAR VMCSs loaded on all cpus if needed.
	 */
	cpu_crash_vmclear_loaded_vmcss();

	/* Disable VMX or SVM if needed.
	 *
	 * We need to disable virtualization on all CPUs.
	 * Having VMX or SVM enabled on any CPU may break rebooting
	 * after the kdump kernel has finished its task.
	 */
	cpu_emergency_vmxoff();
	cpu_emergency_svm_disable();

	/*
	 * Disable Intel PT to stop its logging
	 */
	cpu_emergency_stop_pt();

	disable_local_APIC();
}

static void kdump_nmi_shootdown_cpus(void)
{
	nmi_shootdown_cpus(kdump_nmi_callback);

	disable_local_APIC();
}

#else
static void kdump_nmi_shootdown_cpus(void)
{
	/* There are no cpus to shootdown */
}
#endif

void native_machine_crash_shutdown(struct pt_regs *regs)
{
	/* This function is only called after the system
	 * has panicked or is otherwise in a critical state.
	 * The minimum amount of code to allow a kexec'd kernel
	 * to run successfully needs to happen here.
	 *
	 * In practice this means shooting down the other cpus in
	 * an SMP system.
	 */
	/* The kernel is broken so disable interrupts */
	local_irq_disable();

	kdump_nmi_shootdown_cpus();

	/*
	 * VMCLEAR VMCSs loaded on this cpu if needed.
	 */
	cpu_crash_vmclear_loaded_vmcss();

	/* Booting kdump kernel with VMX or SVM enabled won't work,
	 * because (among other limitations) we can't disable paging
	 * with the virt flags.
	 */
	cpu_emergency_vmxoff();
	cpu_emergency_svm_disable();

	/*
	 * Disable Intel PT to stop its logging
	 */
	cpu_emergency_stop_pt();

#ifdef CONFIG_X86_IO_APIC
	/* Prevent crash_kexec() from deadlocking on ioapic_lock. */
	ioapic_zap_locks();
	disable_IO_APIC();
#endif
	lapic_shutdown();
#ifdef CONFIG_HPET_TIMER
	hpet_disable();
#endif
	crash_save_cpu(regs, safe_smp_processor_id());
}

#ifdef CONFIG_KEXEC_FILE
static int get_nr_ram_ranges_callback(u64 start, u64 end, void *arg)
{
	unsigned int *nr_ranges = arg;

	(*nr_ranges)++;
	return 0;
}

static int get_gart_ranges_callback(u64 start, u64 end, void *arg)
{
	struct crash_elf_data *ced = arg;

	ced->gart_start = start;
	ced->gart_end = end;

	/* Not expecting more than 1 gart aperture */
	return 1;
}


/* Gather all the required information to prepare elf headers for ram regions */
static void fill_up_crash_elf_data(struct crash_elf_data *ced,
				   struct kimage *image)
{
	unsigned int nr_ranges = 0;

	ced->image = image;

	walk_system_ram_res(0, -1, &nr_ranges,
				get_nr_ram_ranges_callback);

	ced->max_nr_ranges = nr_ranges;

	/*
	 * We don't create ELF headers for GART aperture as an attempt
	 * to dump this memory in second kernel leads to hang/crash.
	 * If gart aperture is present, one needs to exclude that region
	 * and that could lead to need of extra phdr.
	 */
	walk_iomem_res("GART", IORESOURCE_MEM, 0, -1,
				ced, get_gart_ranges_callback);

	/*
	 * If we have gart region, excluding that could potentially split
	 * a memory range, resulting in extra header. Account for  that.
	 */
	if (ced->gart_end)
		ced->max_nr_ranges++;

	/* Exclusion of crash region could split memory ranges */
	ced->max_nr_ranges++;

	/* If crashk_low_res is not 0, another range split possible */
	if (crashk_low_res.end)
		ced->max_nr_ranges++;
}

static int exclude_mem_range(struct crash_mem *mem,
		unsigned long long mstart, unsigned long long mend)
{
	int i, j;
	unsigned long long start, end;
	struct crash_mem_range temp_range = {0, 0};

	for (i = 0; i < mem->nr_ranges; i++) {
		start = mem->ranges[i].start;
		end = mem->ranges[i].end;

		if (mstart > end || mend < start)
			continue;

		/* Truncate any area outside of range */
		if (mstart < start)
			mstart = start;
		if (mend > end)
			mend = end;

		/* Found completely overlapping range */
		if (mstart == start && mend == end) {
			mem->ranges[i].start = 0;
			mem->ranges[i].end = 0;
			if (i < mem->nr_ranges - 1) {
				/* Shift rest of the ranges to left */
				for (j = i; j < mem->nr_ranges - 1; j++) {
					mem->ranges[j].start =
						mem->ranges[j+1].start;
					mem->ranges[j].end =
							mem->ranges[j+1].end;
				}
			}
			mem->nr_ranges--;
			return 0;
		}

		if (mstart > start && mend < end) {
			/* Split original range */
			mem->ranges[i].end = mstart - 1;
			temp_range.start = mend + 1;
			temp_range.end = end;
		} else if (mstart != start)
			mem->ranges[i].end = mstart - 1;
		else
			mem->ranges[i].start = mend + 1;
		break;
	}

	/* If a split happend, add the split to array */
	if (!temp_range.end)
		return 0;

	/* Split happened */
	if (i == CRASH_MAX_RANGES - 1) {
		pr_err("Too many crash ranges after split\n");
		return -ENOMEM;
	}

	/* Location where new range should go */
	j = i + 1;
	if (j < mem->nr_ranges) {
		/* Move over all ranges one slot towards the end */
		for (i = mem->nr_ranges - 1; i >= j; i--)
			mem->ranges[i + 1] = mem->ranges[i];
	}

	mem->ranges[j].start = temp_range.start;
	mem->ranges[j].end = temp_range.end;
	mem->nr_ranges++;
	return 0;
}

/*
 * Look for any unwanted ranges between mstart, mend and remove them. This
 * might lead to split and split ranges are put in ced->mem.ranges[] array
 */
static int elf_header_exclude_ranges(struct crash_elf_data *ced,
		unsigned long long mstart, unsigned long long mend)
{
	struct crash_mem *cmem = &ced->mem;
	int ret = 0;

	memset(cmem->ranges, 0, sizeof(cmem->ranges));

	cmem->ranges[0].start = mstart;
	cmem->ranges[0].end = mend;
	cmem->nr_ranges = 1;

	/* Exclude crashkernel region */
	ret = exclude_mem_range(cmem, crashk_res.start, crashk_res.end);
	if (ret)
		return ret;

	if (crashk_low_res.end) {
		ret = exclude_mem_range(cmem, crashk_low_res.start, crashk_low_res.end);
		if (ret)
			return ret;
	}

	/* Exclude GART region */
	if (ced->gart_end) {
		ret = exclude_mem_range(cmem, ced->gart_start, ced->gart_end);
		if (ret)
			return ret;
	}

	return ret;
}

static int prepare_elf64_ram_headers_callback(u64 start, u64 end, void *arg)
{
	struct crash_elf_data *ced = arg;
	Elf64_Ehdr *ehdr;
	Elf64_Phdr *phdr;
	unsigned long mstart, mend;
	struct kimage *image = ced->image;
	struct crash_mem *cmem;
	int ret, i;

	ehdr = ced->ehdr;

	/* Exclude unwanted mem ranges */
	ret = elf_header_exclude_ranges(ced, start, end);
	if (ret)
		return ret;

	/* Go through all the ranges in ced->mem.ranges[] and prepare phdr */
	cmem = &ced->mem;

	for (i = 0; i < cmem->nr_ranges; i++) {
		mstart = cmem->ranges[i].start;
		mend = cmem->ranges[i].end;

		phdr = ced->bufp;
		ced->bufp += sizeof(Elf64_Phdr);

		phdr->p_type = PT_LOAD;
		phdr->p_flags = PF_R|PF_W|PF_X;
		phdr->p_offset  = mstart;

		/*
		 * If a range matches backup region, adjust offset to backup
		 * segment.
		 */
		if (mstart == image->arch.backup_src_start &&
		    (mend - mstart + 1) == image->arch.backup_src_sz)
			phdr->p_offset = image->arch.backup_load_addr;

		phdr->p_paddr = mstart;
		phdr->p_vaddr = (unsigned long long) __va(mstart);
		phdr->p_filesz = phdr->p_memsz = mend - mstart + 1;
		phdr->p_align = 0;
		ehdr->e_phnum++;
		pr_debug("Crash PT_LOAD elf header. phdr=%p vaddr=0x%llx, paddr=0x%llx, sz=0x%llx e_phnum=%d p_offset=0x%llx\n",
			phdr, phdr->p_vaddr, phdr->p_paddr, phdr->p_filesz,
			ehdr->e_phnum, phdr->p_offset);
	}

	return ret;
}

static int prepare_elf64_headers(struct crash_elf_data *ced,
		void **addr, unsigned long *sz)
{
	Elf64_Ehdr *ehdr;
	Elf64_Phdr *phdr;
	unsigned long nr_cpus = num_possible_cpus(), nr_phdr, elf_sz;
	unsigned char *buf, *bufp;
	unsigned int cpu;
	unsigned long long notes_addr;
	int ret;

	/* extra phdr for vmcoreinfo elf note */
	nr_phdr = nr_cpus + 1;
	nr_phdr += ced->max_nr_ranges;

	/*
	 * kexec-tools creates an extra PT_LOAD phdr for kernel text mapping
	 * area on x86_64 (ffffffff80000000 - ffffffffa0000000).
	 * I think this is required by tools like gdb. So same physical
	 * memory will be mapped in two elf headers. One will contain kernel
	 * text virtual addresses and other will have __va(physical) addresses.
	 */

	nr_phdr++;
	elf_sz = sizeof(Elf64_Ehdr) + nr_phdr * sizeof(Elf64_Phdr);
	elf_sz = ALIGN(elf_sz, ELF_CORE_HEADER_ALIGN);

	buf = vzalloc(elf_sz);
	if (!buf)
		return -ENOMEM;

	bufp = buf;
	ehdr = (Elf64_Ehdr *)bufp;
	bufp += sizeof(Elf64_Ehdr);
	memcpy(ehdr->e_ident, ELFMAG, SELFMAG);
	ehdr->e_ident[EI_CLASS] = ELFCLASS64;
	ehdr->e_ident[EI_DATA] = ELFDATA2LSB;
	ehdr->e_ident[EI_VERSION] = EV_CURRENT;
	ehdr->e_ident[EI_OSABI] = ELF_OSABI;
	memset(ehdr->e_ident + EI_PAD, 0, EI_NIDENT - EI_PAD);
	ehdr->e_type = ET_CORE;
	ehdr->e_machine = ELF_ARCH;
	ehdr->e_version = EV_CURRENT;
	ehdr->e_phoff = sizeof(Elf64_Ehdr);
	ehdr->e_ehsize = sizeof(Elf64_Ehdr);
	ehdr->e_phentsize = sizeof(Elf64_Phdr);

	/* Prepare one phdr of type PT_NOTE for each present cpu */
	for_each_present_cpu(cpu) {
		phdr = (Elf64_Phdr *)bufp;
		bufp += sizeof(Elf64_Phdr);
		phdr->p_type = PT_NOTE;
		notes_addr = per_cpu_ptr_to_phys(per_cpu_ptr(crash_notes, cpu));
		phdr->p_offset = phdr->p_paddr = notes_addr;
		phdr->p_filesz = phdr->p_memsz = sizeof(note_buf_t);
		(ehdr->e_phnum)++;
	}

	/* Prepare one PT_NOTE header for vmcoreinfo */
	phdr = (Elf64_Phdr *)bufp;
	bufp += sizeof(Elf64_Phdr);
	phdr->p_type = PT_NOTE;
	phdr->p_offset = phdr->p_paddr = paddr_vmcoreinfo_note();
	phdr->p_filesz = phdr->p_memsz = sizeof(vmcoreinfo_note);
	(ehdr->e_phnum)++;

#ifdef CONFIG_X86_64
	/* Prepare PT_LOAD type program header for kernel text region */
	phdr = (Elf64_Phdr *)bufp;
	bufp += sizeof(Elf64_Phdr);
	phdr->p_type = PT_LOAD;
	phdr->p_flags = PF_R|PF_W|PF_X;
	phdr->p_vaddr = (Elf64_Addr)_text;
	phdr->p_filesz = phdr->p_memsz = _end - _text;
	phdr->p_offset = phdr->p_paddr = __pa_symbol(_text);
	(ehdr->e_phnum)++;
#endif

	/* Prepare PT_LOAD headers for system ram chunks. */
	ced->ehdr = ehdr;
	ced->bufp = bufp;
	ret = walk_system_ram_res(0, -1, ced,
			prepare_elf64_ram_headers_callback);
	if (ret < 0)
		return ret;

	*addr = buf;
	*sz = elf_sz;
	return 0;
}

/* Prepare elf headers. Return addr and size */
static int prepare_elf_headers(struct kimage *image, void **addr,
					unsigned long *sz)
{
	struct crash_elf_data *ced;
	int ret;

	ced = kzalloc(sizeof(*ced), GFP_KERNEL);
	if (!ced)
		return -ENOMEM;

	fill_up_crash_elf_data(ced, image);

	/* By default prepare 64bit headers */
	ret =  prepare_elf64_headers(ced, addr, sz);
	kfree(ced);
	return ret;
}

static int add_e820_entry(struct boot_params *params, struct e820entry *entry)
{
	unsigned int nr_e820_entries;

	nr_e820_entries = params->e820_entries;
	if (nr_e820_entries >= E820MAX)
		return 1;

	memcpy(&params->e820_map[nr_e820_entries], entry,
			sizeof(struct e820entry));
	params->e820_entries++;
	return 0;
}

static int memmap_entry_callback(u64 start, u64 end, void *arg)
{
	struct crash_memmap_data *cmd = arg;
	struct boot_params *params = cmd->params;
	struct e820entry ei;

	ei.addr = start;
	ei.size = end - start + 1;
	ei.type = cmd->type;
	add_e820_entry(params, &ei);

	return 0;
}

static int memmap_exclude_ranges(struct kimage *image, struct crash_mem *cmem,
				 unsigned long long mstart,
				 unsigned long long mend)
{
	unsigned long start, end;
	int ret = 0;

	cmem->ranges[0].start = mstart;
	cmem->ranges[0].end = mend;
	cmem->nr_ranges = 1;

	/* Exclude Backup region */
	start = image->arch.backup_load_addr;
	end = start + image->arch.backup_src_sz - 1;
	ret = exclude_mem_range(cmem, start, end);
	if (ret)
		return ret;

	/* Exclude elf header region */
	start = image->arch.elf_load_addr;
	end = start + image->arch.elf_headers_sz - 1;
	return exclude_mem_range(cmem, start, end);
}

/* Prepare memory map for crash dump kernel */
int crash_setup_memmap_entries(struct kimage *image, struct boot_params *params)
{
	int i, ret = 0;
	unsigned long flags;
	struct e820entry ei;
	struct crash_memmap_data cmd;
	struct crash_mem *cmem;

	cmem = vzalloc(sizeof(struct crash_mem));
	if (!cmem)
		return -ENOMEM;

	memset(&cmd, 0, sizeof(struct crash_memmap_data));
	cmd.params = params;

	/* Add first 640K segment */
	ei.addr = image->arch.backup_src_start;
	ei.size = image->arch.backup_src_sz;
	ei.type = E820_RAM;
	add_e820_entry(params, &ei);

	/* Add ACPI tables */
	cmd.type = E820_ACPI;
	flags = IORESOURCE_MEM | IORESOURCE_BUSY;
	walk_iomem_res("ACPI Tables", flags, 0, -1, &cmd,
		       memmap_entry_callback);

	/* Add ACPI Non-volatile Storage */
	cmd.type = E820_NVS;
	walk_iomem_res("ACPI Non-volatile Storage", flags, 0, -1, &cmd,
			memmap_entry_callback);

	/* Add crashk_low_res region */
	if (crashk_low_res.end) {
		ei.addr = crashk_low_res.start;
		ei.size = crashk_low_res.end - crashk_low_res.start + 1;
		ei.type = E820_RAM;
		add_e820_entry(params, &ei);
	}

	/* Exclude some ranges from crashk_res and add rest to memmap */
	ret = memmap_exclude_ranges(image, cmem, crashk_res.start,
						crashk_res.end);
	if (ret)
		goto out;

	for (i = 0; i < cmem->nr_ranges; i++) {
		ei.size = cmem->ranges[i].end - cmem->ranges[i].start + 1;

		/* If entry is less than a page, skip it */
		if (ei.size < PAGE_SIZE)
			continue;
		ei.addr = cmem->ranges[i].start;
		ei.type = E820_RAM;
		add_e820_entry(params, &ei);
	}

out:
	vfree(cmem);
	return ret;
}

static int determine_backup_region(u64 start, u64 end, void *arg)
{
	struct kimage *image = arg;

	image->arch.backup_src_start = start;
	image->arch.backup_src_sz = end - start + 1;

	/* Expecting only one range for backup region */
	return 1;
}

int crash_load_segments(struct kimage *image)
{
	unsigned long src_start, src_sz, elf_sz;
	void *elf_addr;
	int ret;

	/*
	 * Determine and load a segment for backup area. First 640K RAM
	 * region is backup source
	 */

	ret = walk_system_ram_res(KEXEC_BACKUP_SRC_START, KEXEC_BACKUP_SRC_END,
				image, determine_backup_region);

	/* Zero or postive return values are ok */
	if (ret < 0)
		return ret;

	src_start = image->arch.backup_src_start;
	src_sz = image->arch.backup_src_sz;

	/* Add backup segment. */
	if (src_sz) {
		/*
		 * Ideally there is no source for backup segment. This is
		 * copied in purgatory after crash. Just add a zero filled
		 * segment for now to make sure checksum logic works fine.
		 */
		ret = kexec_add_buffer(image, (char *)&crash_zero_bytes,
				       sizeof(crash_zero_bytes), src_sz,
				       PAGE_SIZE, 0, -1, 0,
				       &image->arch.backup_load_addr);
		if (ret)
			return ret;
		pr_debug("Loaded backup region at 0x%lx backup_start=0x%lx memsz=0x%lx\n",
			 image->arch.backup_load_addr, src_start, src_sz);
	}

	/* Prepare elf headers and add a segment */
	ret = prepare_elf_headers(image, &elf_addr, &elf_sz);
	if (ret)
		return ret;

	image->arch.elf_headers = elf_addr;
	image->arch.elf_headers_sz = elf_sz;

	ret = kexec_add_buffer(image, (char *)elf_addr, elf_sz, elf_sz,
			ELF_CORE_HEADER_ALIGN, 0, -1, 0,
			&image->arch.elf_load_addr);
	if (ret) {
		vfree((void *)image->arch.elf_headers);
		return ret;
	}
	pr_debug("Loaded ELF headers at 0x%lx bufsz=0x%lx memsz=0x%lx\n",
		 image->arch.elf_load_addr, elf_sz, elf_sz);

	return ret;
}
#endif /* CONFIG_KEXEC_FILE */
Commit	Line	Data
5033cba0	1	/*
62a31a03	2	* Architecture specific (i386/x86_64) functions for kexec based crash dumps.
5033cba0 EB	3	*
	4	* Created by: Hariprasad Nellitheertha (hari@in.ibm.com)
	5	*
	6	* Copyright (C) IBM Corporation, 2004. All rights reserved.
dd5f7260 VG	7	* Copyright (C) Red Hat Inc., 2014. All rights reserved.
	8	* Authors:
	9	* Vivek Goyal <vgoyal@redhat.com>
5033cba0 EB	10	*
	11	*/
	12
dd5f7260 VG	13	#define pr_fmt(fmt) "kexec: " fmt
dd5f7260 VG	14
5033cba0 EB	15	#include <linux/types.h>
	16	#include <linux/kernel.h>
	17	#include <linux/smp.h>
5033cba0 EB	18	#include <linux/reboot.h>
5033cba0 EB	19	#include <linux/kexec.h>
5033cba0 EB	20	#include <linux/delay.h>
	21	#include <linux/elf.h>
	22	#include <linux/elfcore.h>
f23d1f4a	23	#include <linux/module.h>
dd5f7260	24	#include <linux/slab.h>
d6472302	25	#include <linux/vmalloc.h>
5033cba0 EB	26
	27	#include <asm/processor.h>
	28	#include <asm/hardirq.h>
	29	#include <asm/nmi.h>
	30	#include <asm/hw_irq.h>
19842d67	31	#include <asm/apic.h>
8643e28d	32	#include <asm/io_apic.h>
0c1b2724	33	#include <asm/hpet.h>
1eeb66a1	34	#include <linux/kdebug.h>
96b89dc6	35	#include <asm/cpu.h>
ed23dc6f	36	#include <asm/reboot.h>
2340b62f	37	#include <asm/virtext.h>
da06a43d	38	#include <asm/intel_pt.h>
8e294786	39
dd5f7260 VG	40	/* Alignment required for elf header segment */
	41	#define ELF_CORE_HEADER_ALIGN 4096
	42
	43	/* This primarily represents number of split ranges due to exclusion */
	44	#define CRASH_MAX_RANGES 16
	45
	46	struct crash_mem_range {
	47	u64 start, end;
	48	};
	49
	50	struct crash_mem {
	51	unsigned int nr_ranges;
	52	struct crash_mem_range ranges[CRASH_MAX_RANGES];
	53	};
	54
	55	/* Misc data about ram ranges needed to prepare elf headers */
	56	struct crash_elf_data {
	57	struct kimage *image;
	58	/*
	59	* Total number of ram ranges we have after various adjustments for
	60	* GART, crash reserved region etc.
	61	*/
	62	unsigned int max_nr_ranges;
	63	unsigned long gart_start, gart_end;
	64
	65	/* Pointer to elf header */
	66	void *ehdr;
	67	/* Pointer to next phdr */
	68	void *bufp;
	69	struct crash_mem mem;
	70	};
	71
	72	/* Used while preparing memory map entries for second kernel */
	73	struct crash_memmap_data {
	74	struct boot_params *params;
	75	/* Type of memory */
	76	unsigned int type;
	77	};
	78
f23d1f4a ZY	79	/*
	80	* This is used to VMCLEAR all VMCSs loaded on the
	81	* processor. And when loading kvm_intel module, the
	82	* callback function pointer will be assigned.
	83	*
	84	* protected by rcu.
	85	*/
0ca0d818	86	crash_vmclear_fn __rcu *crash_vmclear_loaded_vmcss = NULL;
f23d1f4a	87	EXPORT_SYMBOL_GPL(crash_vmclear_loaded_vmcss);
dd5f7260	88	unsigned long crash_zero_bytes;
f23d1f4a ZY	89
	90	static inline void cpu_crash_vmclear_loaded_vmcss(void)
	91	{
0ca0d818	92	crash_vmclear_fn *do_vmclear_operation = NULL;
f23d1f4a ZY	93
	94	rcu_read_lock();
	95	do_vmclear_operation = rcu_dereference(crash_vmclear_loaded_vmcss);
	96	if (do_vmclear_operation)
	97	do_vmclear_operation();
	98	rcu_read_unlock();
	99	}
	100
b2bbe71b EH	101	#if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC)
b2bbe71b EH	102
9c48f1c6	103	static void kdump_nmi_callback(int cpu, struct pt_regs *regs)
c4ac4263	104	{
1fb473d8	105	#ifdef CONFIG_X86_32
4d55476c	106	struct pt_regs fixed_regs;
a7d41820	107
f39b6f0e	108	if (!user_mode(regs)) {
a7d41820 EH	109	crash_fixup_ss_esp(&fixed_regs, regs);
	110	regs = &fixed_regs;
	111	}
	112	#endif
	113	crash_save_cpu(regs, cpu);
	114
f23d1f4a ZY	115	/*
	116	* VMCLEAR VMCSs loaded on all cpus if needed.
	117	*/
	118	cpu_crash_vmclear_loaded_vmcss();
	119
2340b62f EH	120	/* Disable VMX or SVM if needed.
	121	*
	122	* We need to disable virtualization on all CPUs.
	123	* Having VMX or SVM enabled on any CPU may break rebooting
	124	* after the kdump kernel has finished its task.
	125	*/
	126	cpu_emergency_vmxoff();
	127	cpu_emergency_svm_disable();
	128
da06a43d TI	129	/*
	130	* Disable Intel PT to stop its logging
	131	*/
	132	cpu_emergency_stop_pt();
	133
a7d41820 EH	134	disable_local_APIC();
	135	}
	136
d1e7b91c EH	137	static void kdump_nmi_shootdown_cpus(void)
d1e7b91c EH	138	{
8e294786	139	nmi_shootdown_cpus(kdump_nmi_callback);
d1e7b91c	140
19842d67	141	disable_local_APIC();
c4ac4263	142	}
d1e7b91c	143
c4ac4263	144	#else
d1e7b91c	145	static void kdump_nmi_shootdown_cpus(void)
c4ac4263 EB	146	{
	147	/* There are no cpus to shootdown */
	148	}
	149	#endif
	150
ed23dc6f	151	void native_machine_crash_shutdown(struct pt_regs *regs)
5033cba0 EB	152	{
5033cba0 EB	153	/* This function is only called after the system
f18190bd	154	* has panicked or is otherwise in a critical state.
5033cba0 EB	155	* The minimum amount of code to allow a kexec'd kernel
	156	* to run successfully needs to happen here.
	157	*
	158	* In practice this means shooting down the other cpus in
	159	* an SMP system.
	160	*/
c4ac4263 EB	161	/* The kernel is broken so disable interrupts */
c4ac4263 EB	162	local_irq_disable();
a3ea8ac8	163
d1e7b91c	164	kdump_nmi_shootdown_cpus();
2340b62f	165
f23d1f4a ZY	166	/*
	167	* VMCLEAR VMCSs loaded on this cpu if needed.
	168	*/
	169	cpu_crash_vmclear_loaded_vmcss();
	170
2340b62f EH	171	/* Booting kdump kernel with VMX or SVM enabled won't work,
	172	* because (among other limitations) we can't disable paging
	173	* with the virt flags.
	174	*/
	175	cpu_emergency_vmxoff();
	176	cpu_emergency_svm_disable();
	177
da06a43d TI	178	/*
	179	* Disable Intel PT to stop its logging
	180	*/
	181	cpu_emergency_stop_pt();
	182
17405453 YY	183	#ifdef CONFIG_X86_IO_APIC
	184	/* Prevent crash_kexec() from deadlocking on ioapic_lock. */
	185	ioapic_zap_locks();
19842d67	186	disable_IO_APIC();
0c1b2724	187	#endif
522e6646	188	lapic_shutdown();
0c1b2724 OH	189	#ifdef CONFIG_HPET_TIMER
0c1b2724 OH	190	hpet_disable();
19842d67	191	#endif
85916f81	192	crash_save_cpu(regs, safe_smp_processor_id());
5033cba0	193	}
dd5f7260	194
74ca317c	195	#ifdef CONFIG_KEXEC_FILE
e3c41e37	196	static int get_nr_ram_ranges_callback(u64 start, u64 end, void *arg)
dd5f7260	197	{
e3c41e37	198	unsigned int *nr_ranges = arg;
dd5f7260 VG	199
	200	(*nr_ranges)++;
	201	return 0;
	202	}
	203
	204	static int get_gart_ranges_callback(u64 start, u64 end, void *arg)
	205	{
	206	struct crash_elf_data *ced = arg;
	207
	208	ced->gart_start = start;
	209	ced->gart_end = end;
	210
	211	/* Not expecting more than 1 gart aperture */
	212	return 1;
	213	}
	214
	215
	216	/* Gather all the required information to prepare elf headers for ram regions */
	217	static void fill_up_crash_elf_data(struct crash_elf_data *ced,
	218	struct kimage *image)
	219	{
	220	unsigned int nr_ranges = 0;
	221
	222	ced->image = image;
	223
e3c41e37	224	walk_system_ram_res(0, -1, &nr_ranges,
dd5f7260 VG	225	get_nr_ram_ranges_callback);
	226
	227	ced->max_nr_ranges = nr_ranges;
	228
	229	/*
	230	* We don't create ELF headers for GART aperture as an attempt
	231	* to dump this memory in second kernel leads to hang/crash.
	232	* If gart aperture is present, one needs to exclude that region
	233	* and that could lead to need of extra phdr.
	234	*/
	235	walk_iomem_res("GART", IORESOURCE_MEM, 0, -1,
	236	ced, get_gart_ranges_callback);
	237
	238	/*
	239	* If we have gart region, excluding that could potentially split
	240	* a memory range, resulting in extra header. Account for that.
	241	*/
	242	if (ced->gart_end)
	243	ced->max_nr_ranges++;
	244
	245	/* Exclusion of crash region could split memory ranges */
	246	ced->max_nr_ranges++;
	247
	248	/* If crashk_low_res is not 0, another range split possible */
a2d6aa8f	249	if (crashk_low_res.end)
dd5f7260 VG	250	ced->max_nr_ranges++;
	251	}
	252
	253	static int exclude_mem_range(struct crash_mem *mem,
	254	unsigned long long mstart, unsigned long long mend)
	255	{
	256	int i, j;
	257	unsigned long long start, end;
	258	struct crash_mem_range temp_range = {0, 0};
	259
	260	for (i = 0; i < mem->nr_ranges; i++) {
	261	start = mem->ranges[i].start;
	262	end = mem->ranges[i].end;
	263
	264	if (mstart > end \|\| mend < start)
	265	continue;
	266
	267	/* Truncate any area outside of range */
	268	if (mstart < start)
	269	mstart = start;
	270	if (mend > end)
	271	mend = end;
	272
	273	/* Found completely overlapping range */
	274	if (mstart == start && mend == end) {
	275	mem->ranges[i].start = 0;
	276	mem->ranges[i].end = 0;
	277	if (i < mem->nr_ranges - 1) {
	278	/* Shift rest of the ranges to left */
	279	for (j = i; j < mem->nr_ranges - 1; j++) {
	280	mem->ranges[j].start =
	281	mem->ranges[j+1].start;
	282	mem->ranges[j].end =
	283	mem->ranges[j+1].end;
	284	}
	285	}
	286	mem->nr_ranges--;
	287	return 0;
	288	}
	289
	290	if (mstart > start && mend < end) {
	291	/* Split original range */
	292	mem->ranges[i].end = mstart - 1;
	293	temp_range.start = mend + 1;
	294	temp_range.end = end;
	295	} else if (mstart != start)
	296	mem->ranges[i].end = mstart - 1;
	297	else
	298	mem->ranges[i].start = mend + 1;
	299	break;
	300	}
	301
	302	/* If a split happend, add the split to array */
	303	if (!temp_range.end)
	304	return 0;
	305
	306	/* Split happened */
	307	if (i == CRASH_MAX_RANGES - 1) {
	308	pr_err("Too many crash ranges after split\n");
	309	return -ENOMEM;
	310	}
	311
	312	/* Location where new range should go */
	313	j = i + 1;
314	if (j < mem->nr_ranges) {
315	/* Move over all ranges one slot towards the end */
316	for (i = mem->nr_ranges - 1; i >= j; i--)
317	mem->ranges[i + 1] = mem->ranges[i];
318	}
319
320	mem->ranges[j].start = temp_range.start;
321	mem->ranges[j].end = temp_range.end;
322	mem->nr_ranges++;
323	return 0;
324	}
325
326	/*
327	* Look for any unwanted ranges between mstart, mend and remove them. This
328	* might lead to split and split ranges are put in ced->mem.ranges[] array
329	*/
330	static int elf_header_exclude_ranges(struct crash_elf_data *ced,
331	unsigned long long mstart, unsigned long long mend)
332	{
333	struct crash_mem *cmem = &ced->mem;
334	int ret = 0;
335
336	memset(cmem->ranges, 0, sizeof(cmem->ranges));
337
338	cmem->ranges[0].start = mstart;
339	cmem->ranges[0].end = mend;
340	cmem->nr_ranges = 1;
341
342	/* Exclude crashkernel region */
343	ret = exclude_mem_range(cmem, crashk_res.start, crashk_res.end);
344	if (ret)
345	return ret;
346
a2d6aa8f BH	347	if (crashk_low_res.end) {
	348	ret = exclude_mem_range(cmem, crashk_low_res.start, crashk_low_res.end);
	349	if (ret)
	350	return ret;
	351	}
dd5f7260 VG	352
	353	/* Exclude GART region */
	354	if (ced->gart_end) {
	355	ret = exclude_mem_range(cmem, ced->gart_start, ced->gart_end);
	356	if (ret)
	357	return ret;
	358	}
	359
	360	return ret;
	361	}
	362
	363	static int prepare_elf64_ram_headers_callback(u64 start, u64 end, void *arg)
	364	{
	365	struct crash_elf_data *ced = arg;
	366	Elf64_Ehdr *ehdr;
	367	Elf64_Phdr *phdr;
	368	unsigned long mstart, mend;
	369	struct kimage *image = ced->image;
	370	struct crash_mem *cmem;
	371	int ret, i;
	372
	373	ehdr = ced->ehdr;
	374
	375	/* Exclude unwanted mem ranges */
	376	ret = elf_header_exclude_ranges(ced, start, end);
	377	if (ret)
	378	return ret;
	379
	380	/* Go through all the ranges in ced->mem.ranges[] and prepare phdr */
	381	cmem = &ced->mem;
	382
	383	for (i = 0; i < cmem->nr_ranges; i++) {
	384	mstart = cmem->ranges[i].start;
	385	mend = cmem->ranges[i].end;
	386
	387	phdr = ced->bufp;
	388	ced->bufp += sizeof(Elf64_Phdr);
	389
	390	phdr->p_type = PT_LOAD;
	391	phdr->p_flags = PF_R\|PF_W\|PF_X;
	392	phdr->p_offset = mstart;
	393
	394	/*
	395	* If a range matches backup region, adjust offset to backup
	396	* segment.
	397	*/
	398	if (mstart == image->arch.backup_src_start &&
	399	(mend - mstart + 1) == image->arch.backup_src_sz)
	400	phdr->p_offset = image->arch.backup_load_addr;
	401
	402	phdr->p_paddr = mstart;
	403	phdr->p_vaddr = (unsigned long long) __va(mstart);
	404	phdr->p_filesz = phdr->p_memsz = mend - mstart + 1;
	405	phdr->p_align = 0;
	406	ehdr->e_phnum++;
	407	pr_debug("Crash PT_LOAD elf header. phdr=%p vaddr=0x%llx, paddr=0x%llx, sz=0x%llx e_phnum=%d p_offset=0x%llx\n",
	408	phdr, phdr->p_vaddr, phdr->p_paddr, phdr->p_filesz,
	409	ehdr->e_phnum, phdr->p_offset);
	410	}
	411
	412	return ret;
	413	}
	414
	415	static int prepare_elf64_headers(struct crash_elf_data *ced,
416	void *addr, unsigned long sz)
417	{
418	Elf64_Ehdr *ehdr;
419	Elf64_Phdr *phdr;
420	unsigned long nr_cpus = num_possible_cpus(), nr_phdr, elf_sz;
421	unsigned char buf, bufp;
422	unsigned int cpu;
423	unsigned long long notes_addr;
424	int ret;
425
426	/* extra phdr for vmcoreinfo elf note */
427	nr_phdr = nr_cpus + 1;
428	nr_phdr += ced->max_nr_ranges;
429
430	/*
431	* kexec-tools creates an extra PT_LOAD phdr for kernel text mapping
432	* area on x86_64 (ffffffff80000000 - ffffffffa0000000).
433	* I think this is required by tools like gdb. So same physical
434	* memory will be mapped in two elf headers. One will contain kernel
435	* text virtual addresses and other will have __va(physical) addresses.
436	*/
437
438	nr_phdr++;
439	elf_sz = sizeof(Elf64_Ehdr) + nr_phdr * sizeof(Elf64_Phdr);
440	elf_sz = ALIGN(elf_sz, ELF_CORE_HEADER_ALIGN);
441
442	buf = vzalloc(elf_sz);
443	if (!buf)
444	return -ENOMEM;
445
446	bufp = buf;
447	ehdr = (Elf64_Ehdr *)bufp;
448	bufp += sizeof(Elf64_Ehdr);
449	memcpy(ehdr->e_ident, ELFMAG, SELFMAG);
450	ehdr->e_ident[EI_CLASS] = ELFCLASS64;
451	ehdr->e_ident[EI_DATA] = ELFDATA2LSB;
452	ehdr->e_ident[EI_VERSION] = EV_CURRENT;
453	ehdr->e_ident[EI_OSABI] = ELF_OSABI;
454	memset(ehdr->e_ident + EI_PAD, 0, EI_NIDENT - EI_PAD);
455	ehdr->e_type = ET_CORE;
456	ehdr->e_machine = ELF_ARCH;
457	ehdr->e_version = EV_CURRENT;
458	ehdr->e_phoff = sizeof(Elf64_Ehdr);
459	ehdr->e_ehsize = sizeof(Elf64_Ehdr);
460	ehdr->e_phentsize = sizeof(Elf64_Phdr);
461
462	/* Prepare one phdr of type PT_NOTE for each present cpu */
463	for_each_present_cpu(cpu) {
464	phdr = (Elf64_Phdr *)bufp;
465	bufp += sizeof(Elf64_Phdr);
466	phdr->p_type = PT_NOTE;
467	notes_addr = per_cpu_ptr_to_phys(per_cpu_ptr(crash_notes, cpu));
468	phdr->p_offset = phdr->p_paddr = notes_addr;
469	phdr->p_filesz = phdr->p_memsz = sizeof(note_buf_t);
470	(ehdr->e_phnum)++;
471	}
472
473	/* Prepare one PT_NOTE header for vmcoreinfo */
474	phdr = (Elf64_Phdr *)bufp;
475	bufp += sizeof(Elf64_Phdr);
476	phdr->p_type = PT_NOTE;
477	phdr->p_offset = phdr->p_paddr = paddr_vmcoreinfo_note();
478	phdr->p_filesz = phdr->p_memsz = sizeof(vmcoreinfo_note);
479	(ehdr->e_phnum)++;
480
481	#ifdef CONFIG_X86_64
482	/* Prepare PT_LOAD type program header for kernel text region */
483	phdr = (Elf64_Phdr *)bufp;
484	bufp += sizeof(Elf64_Phdr);
485	phdr->p_type = PT_LOAD;
486	phdr->p_flags = PF_R\|PF_W\|PF_X;
487	phdr->p_vaddr = (Elf64_Addr)_text;
488	phdr->p_filesz = phdr->p_memsz = _end - _text;
489	phdr->p_offset = phdr->p_paddr = __pa_symbol(_text);
490	(ehdr->e_phnum)++;
491	#endif
492
493	/* Prepare PT_LOAD headers for system ram chunks. */
494	ced->ehdr = ehdr;
495	ced->bufp = bufp;
496	ret = walk_system_ram_res(0, -1, ced,
497	prepare_elf64_ram_headers_callback);
498	if (ret < 0)
499	return ret;
500
501	*addr = buf;
502	*sz = elf_sz;
503	return 0;
504	}
505
506	/* Prepare elf headers. Return addr and size */
507	static int prepare_elf_headers(struct kimage image, void *addr,
508	unsigned long *sz)
509	{
510	struct crash_elf_data *ced;
511	int ret;
512
513	ced = kzalloc(sizeof(*ced), GFP_KERNEL);
514	if (!ced)
515	return -ENOMEM;
516
517	fill_up_crash_elf_data(ced, image);
518
519	/* By default prepare 64bit headers */
520	ret = prepare_elf64_headers(ced, addr, sz);
521	kfree(ced);
522	return ret;
523	}
524
525	static int add_e820_entry(struct boot_params params, struct e820entry entry)
526	{
527	unsigned int nr_e820_entries;
528
529	nr_e820_entries = params->e820_entries;
530	if (nr_e820_entries >= E820MAX)
531	return 1;
532
533	memcpy(&params->e820_map[nr_e820_entries], entry,
534	sizeof(struct e820entry));
535	params->e820_entries++;
536	return 0;
537	}
538
539	static int memmap_entry_callback(u64 start, u64 end, void *arg)
540	{
541	struct crash_memmap_data *cmd = arg;
542	struct boot_params *params = cmd->params;
543	struct e820entry ei;
544
545	ei.addr = start;
546	ei.size = end - start + 1;
547	ei.type = cmd->type;
548	add_e820_entry(params, &ei);
549
550	return 0;
551	}
552
553	static int memmap_exclude_ranges(struct kimage image, struct crash_mem cmem,
554	unsigned long long mstart,
555	unsigned long long mend)
556	{
557	unsigned long start, end;
558	int ret = 0;
559
560	cmem->ranges[0].start = mstart;
561	cmem->ranges[0].end = mend;
562	cmem->nr_ranges = 1;
563
564	/* Exclude Backup region */
565	start = image->arch.backup_load_addr;
566	end = start + image->arch.backup_src_sz - 1;
567	ret = exclude_mem_range(cmem, start, end);
568	if (ret)
569	return ret;
570
571	/* Exclude elf header region */
572	start = image->arch.elf_load_addr;
573	end = start + image->arch.elf_headers_sz - 1;
574	return exclude_mem_range(cmem, start, end);
575	}
576
577	/* Prepare memory map for crash dump kernel */
578	int crash_setup_memmap_entries(struct kimage image, struct boot_params params)
579	{
580	int i, ret = 0;
581	unsigned long flags;
582	struct e820entry ei;
583	struct crash_memmap_data cmd;
584	struct crash_mem *cmem;
585
586	cmem = vzalloc(sizeof(struct crash_mem));
587	if (!cmem)
588	return -ENOMEM;
589
590	memset(&cmd, 0, sizeof(struct crash_memmap_data));
591	cmd.params = params;
592
593	/* Add first 640K segment */
594	ei.addr = image->arch.backup_src_start;
595	ei.size = image->arch.backup_src_sz;
596	ei.type = E820_RAM;
597	add_e820_entry(params, &ei);
598
599	/* Add ACPI tables */
600	cmd.type = E820_ACPI;
601	flags = IORESOURCE_MEM \| IORESOURCE_BUSY;
602	walk_iomem_res("ACPI Tables", flags, 0, -1, &cmd,
603	memmap_entry_callback);
604
605	/* Add ACPI Non-volatile Storage */
606	cmd.type = E820_NVS;
607	walk_iomem_res("ACPI Non-volatile Storage", flags, 0, -1, &cmd,
608	memmap_entry_callback);
609
610	/* Add crashk_low_res region */
611	if (crashk_low_res.end) {
612	ei.addr = crashk_low_res.start;
613	ei.size = crashk_low_res.end - crashk_low_res.start + 1;
614	ei.type = E820_RAM;
615	add_e820_entry(params, &ei);
616	}
617
618	/* Exclude some ranges from crashk_res and add rest to memmap */
619	ret = memmap_exclude_ranges(image, cmem, crashk_res.start,
620	crashk_res.end);
621	if (ret)
622	goto out;
623
624	for (i = 0; i < cmem->nr_ranges; i++) {
625	ei.size = cmem->ranges[i].end - cmem->ranges[i].start + 1;
626
627	/* If entry is less than a page, skip it */
628	if (ei.size < PAGE_SIZE)
629	continue;
630	ei.addr = cmem->ranges[i].start;
631	ei.type = E820_RAM;
632	add_e820_entry(params, &ei);
633	}
634
635	out:
636	vfree(cmem);
637	return ret;
638	}
639
640	static int determine_backup_region(u64 start, u64 end, void *arg)
641	{
642	struct kimage *image = arg;
643
644	image->arch.backup_src_start = start;
645	image->arch.backup_src_sz = end - start + 1;
646
647	/* Expecting only one range for backup region */
648	return 1;
649	}
650
651	int crash_load_segments(struct kimage *image)
652	{
653	unsigned long src_start, src_sz, elf_sz;
654	void *elf_addr;
655	int ret;
656
657	/*
658	* Determine and load a segment for backup area. First 640K RAM
659	* region is backup source
660	*/
661
662	ret = walk_system_ram_res(KEXEC_BACKUP_SRC_START, KEXEC_BACKUP_SRC_END,
663	image, determine_backup_region);
664
665	/* Zero or postive return values are ok */
666	if (ret < 0)
667	return ret;
668
669	src_start = image->arch.backup_src_start;
670	src_sz = image->arch.backup_src_sz;
671
672	/* Add backup segment. */
673	if (src_sz) {
674	/*
675	* Ideally there is no source for backup segment. This is
676	* copied in purgatory after crash. Just add a zero filled
677	* segment for now to make sure checksum logic works fine.
678	*/
679	ret = kexec_add_buffer(image, (char *)&crash_zero_bytes,
680	sizeof(crash_zero_bytes), src_sz,
681	PAGE_SIZE, 0, -1, 0,
682	&image->arch.backup_load_addr);
683	if (ret)
684	return ret;
685	pr_debug("Loaded backup region at 0x%lx backup_start=0x%lx memsz=0x%lx\n",
686	image->arch.backup_load_addr, src_start, src_sz);
687	}
688
689	/* Prepare elf headers and add a segment */
690	ret = prepare_elf_headers(image, &elf_addr, &elf_sz);
691	if (ret)
692	return ret;
693
694	image->arch.elf_headers = elf_addr;
695	image->arch.elf_headers_sz = elf_sz;
696
697	ret = kexec_add_buffer(image, (char *)elf_addr, elf_sz, elf_sz,
698	ELF_CORE_HEADER_ALIGN, 0, -1, 0,
699	&image->arch.elf_load_addr);
700	if (ret) {
701	vfree((void *)image->arch.elf_headers);
702	return ret;
703	}
704	pr_debug("Loaded ELF headers at 0x%lx bufsz=0x%lx memsz=0x%lx\n",
705	image->arch.elf_load_addr, elf_sz, elf_sz);
706
707	return ret;
708	}
74ca317c	709	#endif /* CONFIG_KEXEC_FILE */