[linux.git] / fs / proc / kcore.c

// SPDX-License-Identifier: GPL-2.0
/*
 *	fs/proc/kcore.c kernel ELF core dumper
 *
 *	Modelled on fs/exec.c:aout_core_dump()
 *	Jeremy Fitzhardinge <[email protected]>
 *	ELF version written by David Howells <[email protected]>
 *	Modified and incorporated into 2.3.x by Tigran Aivazian <[email protected]>
 *	Support to dump vmalloc'd areas (ELF only), Tigran Aivazian <[email protected]>
 *	Safe accesses to vmalloc/direct-mapped discontiguous areas, Kanoj Sarcar <[email protected]>
 */

#include <linux/crash_core.h>
#include <linux/mm.h>
#include <linux/proc_fs.h>
#include <linux/kcore.h>
#include <linux/user.h>
#include <linux/capability.h>
#include <linux/elf.h>
#include <linux/elfcore.h>
#include <linux/notifier.h>
#include <linux/vmalloc.h>
#include <linux/highmem.h>
#include <linux/printk.h>
#include <linux/bootmem.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
#include <asm/io.h>
#include <linux/list.h>
#include <linux/ioport.h>
#include <linux/memory.h>
#include <linux/sched/task.h>
#include <asm/sections.h>
#include "internal.h"

#define CORE_STR "CORE"

#ifndef ELF_CORE_EFLAGS
#define ELF_CORE_EFLAGS	0
#endif

static struct proc_dir_entry *proc_root_kcore;


#ifndef kc_vaddr_to_offset
#define	kc_vaddr_to_offset(v) ((v) - PAGE_OFFSET)
#endif
#ifndef	kc_offset_to_vaddr
#define	kc_offset_to_vaddr(o) ((o) + PAGE_OFFSET)
#endif

static LIST_HEAD(kclist_head);
static DECLARE_RWSEM(kclist_lock);
static int kcore_need_update = 1;

/* This doesn't grab kclist_lock, so it should only be used at init time. */
void __init kclist_add(struct kcore_list *new, void *addr, size_t size,
		       int type)
{
	new->addr = (unsigned long)addr;
	new->size = size;
	new->type = type;

	list_add_tail(&new->list, &kclist_head);
}

static size_t get_kcore_size(int *nphdr, size_t *phdrs_len, size_t *notes_len,
			     size_t *data_offset)
{
	size_t try, size;
	struct kcore_list *m;

	*nphdr = 1; /* PT_NOTE */
	size = 0;

	list_for_each_entry(m, &kclist_head, list) {
		try = kc_vaddr_to_offset((size_t)m->addr + m->size);
		if (try > size)
			size = try;
		*nphdr = *nphdr + 1;
	}

	*phdrs_len = *nphdr * sizeof(struct elf_phdr);
	*notes_len = (4 * sizeof(struct elf_note) +
		      3 * ALIGN(sizeof(CORE_STR), 4) +
		      VMCOREINFO_NOTE_NAME_BYTES +
		      ALIGN(sizeof(struct elf_prstatus), 4) +
		      ALIGN(sizeof(struct elf_prpsinfo), 4) +
		      ALIGN(arch_task_struct_size, 4) +
		      ALIGN(vmcoreinfo_size, 4));
	*data_offset = PAGE_ALIGN(sizeof(struct elfhdr) + *phdrs_len +
				  *notes_len);
	return *data_offset + size;
}

#ifdef CONFIG_HIGHMEM
/*
 * If no highmem, we can assume [0...max_low_pfn) continuous range of memory
 * because memory hole is not as big as !HIGHMEM case.
 * (HIGHMEM is special because part of memory is _invisible_ from the kernel.)
 */
static int kcore_ram_list(struct list_head *head)
{
	struct kcore_list *ent;

	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
	if (!ent)
		return -ENOMEM;
	ent->addr = (unsigned long)__va(0);
	ent->size = max_low_pfn << PAGE_SHIFT;
	ent->type = KCORE_RAM;
	list_add(&ent->list, head);
	return 0;
}

#else /* !CONFIG_HIGHMEM */

#ifdef CONFIG_SPARSEMEM_VMEMMAP
/* calculate vmemmap's address from given system ram pfn and register it */
static int
get_sparsemem_vmemmap_info(struct kcore_list *ent, struct list_head *head)
{
	unsigned long pfn = __pa(ent->addr) >> PAGE_SHIFT;
	unsigned long nr_pages = ent->size >> PAGE_SHIFT;
	unsigned long start, end;
	struct kcore_list *vmm, *tmp;


	start = ((unsigned long)pfn_to_page(pfn)) & PAGE_MASK;
	end = ((unsigned long)pfn_to_page(pfn + nr_pages)) - 1;
	end = PAGE_ALIGN(end);
	/* overlap check (because we have to align page */
	list_for_each_entry(tmp, head, list) {
		if (tmp->type != KCORE_VMEMMAP)
			continue;
		if (start < tmp->addr + tmp->size)
			if (end > tmp->addr)
				end = tmp->addr;
	}
	if (start < end) {
		vmm = kmalloc(sizeof(*vmm), GFP_KERNEL);
		if (!vmm)
			return 0;
		vmm->addr = start;
		vmm->size = end - start;
		vmm->type = KCORE_VMEMMAP;
		list_add_tail(&vmm->list, head);
	}
	return 1;

}
#else
static int
get_sparsemem_vmemmap_info(struct kcore_list *ent, struct list_head *head)
{
	return 1;
}

#endif

static int
kclist_add_private(unsigned long pfn, unsigned long nr_pages, void *arg)
{
	struct list_head *head = (struct list_head *)arg;
	struct kcore_list *ent;
	struct page *p;

	if (!pfn_valid(pfn))
		return 1;

	p = pfn_to_page(pfn);
	if (!memmap_valid_within(pfn, p, page_zone(p)))
		return 1;

	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
	if (!ent)
		return -ENOMEM;
	ent->addr = (unsigned long)page_to_virt(p);
	ent->size = nr_pages << PAGE_SHIFT;

	if (!virt_addr_valid(ent->addr))
		goto free_out;

	/* cut not-mapped area. ....from ppc-32 code. */
	if (ULONG_MAX - ent->addr < ent->size)
		ent->size = ULONG_MAX - ent->addr;

	/*
	 * We've already checked virt_addr_valid so we know this address
	 * is a valid pointer, therefore we can check against it to determine
	 * if we need to trim
	 */
	if (VMALLOC_START > ent->addr) {
		if (VMALLOC_START - ent->addr < ent->size)
			ent->size = VMALLOC_START - ent->addr;
	}

	ent->type = KCORE_RAM;
	list_add_tail(&ent->list, head);

	if (!get_sparsemem_vmemmap_info(ent, head)) {
		list_del(&ent->list);
		goto free_out;
	}

	return 0;
free_out:
	kfree(ent);
	return 1;
}

static int kcore_ram_list(struct list_head *list)
{
	int nid, ret;
	unsigned long end_pfn;

	/* Not inialized....update now */
	/* find out "max pfn" */
	end_pfn = 0;
	for_each_node_state(nid, N_MEMORY) {
		unsigned long node_end;
		node_end = node_end_pfn(nid);
		if (end_pfn < node_end)
			end_pfn = node_end;
	}
	/* scan 0 to max_pfn */
	ret = walk_system_ram_range(0, end_pfn, list, kclist_add_private);
	if (ret)
		return -ENOMEM;
	return 0;
}
#endif /* CONFIG_HIGHMEM */

static int kcore_update_ram(void)
{
	LIST_HEAD(list);
	LIST_HEAD(garbage);
	int nphdr;
	size_t phdrs_len, notes_len, data_offset;
	struct kcore_list *tmp, *pos;
	int ret = 0;

	down_write(&kclist_lock);
	if (!xchg(&kcore_need_update, 0))
		goto out;

	ret = kcore_ram_list(&list);
	if (ret) {
		/* Couldn't get the RAM list, try again next time. */
		WRITE_ONCE(kcore_need_update, 1);
		list_splice_tail(&list, &garbage);
		goto out;
	}

	list_for_each_entry_safe(pos, tmp, &kclist_head, list) {
		if (pos->type == KCORE_RAM || pos->type == KCORE_VMEMMAP)
			list_move(&pos->list, &garbage);
	}
	list_splice_tail(&list, &kclist_head);

	proc_root_kcore->size = get_kcore_size(&nphdr, &phdrs_len, &notes_len,
					       &data_offset);

out:
	up_write(&kclist_lock);
	list_for_each_entry_safe(pos, tmp, &garbage, list) {
		list_del(&pos->list);
		kfree(pos);
	}
	return ret;
}

static void append_kcore_note(char *notes, size_t *i, const char *name,
			      unsigned int type, const void *desc,
			      size_t descsz)
{
	struct elf_note *note = (struct elf_note *)&notes[*i];

	note->n_namesz = strlen(name) + 1;
	note->n_descsz = descsz;
	note->n_type = type;
	*i += sizeof(*note);
	memcpy(&notes[*i], name, note->n_namesz);
	*i = ALIGN(*i + note->n_namesz, 4);
	memcpy(&notes[*i], desc, descsz);
	*i = ALIGN(*i + descsz, 4);
}

static ssize_t
read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)
{
	char *buf = file->private_data;
	size_t phdrs_offset, notes_offset, data_offset;
	size_t phdrs_len, notes_len;
	struct kcore_list *m;
	size_t tsz;
	int nphdr;
	unsigned long start;
	size_t orig_buflen = buflen;
	int ret = 0;

	down_read(&kclist_lock);

	get_kcore_size(&nphdr, &phdrs_len, &notes_len, &data_offset);
	phdrs_offset = sizeof(struct elfhdr);
	notes_offset = phdrs_offset + phdrs_len;

	/* ELF file header. */
	if (buflen && *fpos < sizeof(struct elfhdr)) {
		struct elfhdr ehdr = {
			.e_ident = {
				[EI_MAG0] = ELFMAG0,
				[EI_MAG1] = ELFMAG1,
				[EI_MAG2] = ELFMAG2,
				[EI_MAG3] = ELFMAG3,
				[EI_CLASS] = ELF_CLASS,
				[EI_DATA] = ELF_DATA,
				[EI_VERSION] = EV_CURRENT,
				[EI_OSABI] = ELF_OSABI,
			},
			.e_type = ET_CORE,
			.e_machine = ELF_ARCH,
			.e_version = EV_CURRENT,
			.e_phoff = sizeof(struct elfhdr),
			.e_flags = ELF_CORE_EFLAGS,
			.e_ehsize = sizeof(struct elfhdr),
			.e_phentsize = sizeof(struct elf_phdr),
			.e_phnum = nphdr,
		};

		tsz = min_t(size_t, buflen, sizeof(struct elfhdr) - *fpos);
		if (copy_to_user(buffer, (char *)&ehdr + *fpos, tsz)) {
			ret = -EFAULT;
			goto out;
		}

		buffer += tsz;
		buflen -= tsz;
		*fpos += tsz;
	}

	/* ELF program headers. */
	if (buflen && *fpos < phdrs_offset + phdrs_len) {
		struct elf_phdr *phdrs, *phdr;

		phdrs = kzalloc(phdrs_len, GFP_KERNEL);
		if (!phdrs) {
			ret = -ENOMEM;
			goto out;
		}

		phdrs[0].p_type = PT_NOTE;
		phdrs[0].p_offset = notes_offset;
		phdrs[0].p_filesz = notes_len;

		phdr = &phdrs[1];
		list_for_each_entry(m, &kclist_head, list) {
			phdr->p_type = PT_LOAD;
			phdr->p_flags = PF_R | PF_W | PF_X;
			phdr->p_offset = kc_vaddr_to_offset(m->addr) + data_offset;
			if (m->type == KCORE_REMAP)
				phdr->p_vaddr = (size_t)m->vaddr;
			else
				phdr->p_vaddr = (size_t)m->addr;
			if (m->type == KCORE_RAM || m->type == KCORE_REMAP)
				phdr->p_paddr = __pa(m->addr);
			else if (m->type == KCORE_TEXT)
				phdr->p_paddr = __pa_symbol(m->addr);
			else
				phdr->p_paddr = (elf_addr_t)-1;
			phdr->p_filesz = phdr->p_memsz = m->size;
			phdr->p_align = PAGE_SIZE;
			phdr++;
		}

		tsz = min_t(size_t, buflen, phdrs_offset + phdrs_len - *fpos);
		if (copy_to_user(buffer, (char *)phdrs + *fpos - phdrs_offset,
				 tsz)) {
			kfree(phdrs);
			ret = -EFAULT;
			goto out;
		}
		kfree(phdrs);

		buffer += tsz;
		buflen -= tsz;
		*fpos += tsz;
	}

	/* ELF note segment. */
	if (buflen && *fpos < notes_offset + notes_len) {
		struct elf_prstatus prstatus = {};
		struct elf_prpsinfo prpsinfo = {
			.pr_sname = 'R',
			.pr_fname = "vmlinux",
		};
		char *notes;
		size_t i = 0;

		strlcpy(prpsinfo.pr_psargs, saved_command_line,
			sizeof(prpsinfo.pr_psargs));

		notes = kzalloc(notes_len, GFP_KERNEL);
		if (!notes) {
			ret = -ENOMEM;
			goto out;
		}

		append_kcore_note(notes, &i, CORE_STR, NT_PRSTATUS, &prstatus,
				  sizeof(prstatus));
		append_kcore_note(notes, &i, CORE_STR, NT_PRPSINFO, &prpsinfo,
				  sizeof(prpsinfo));
		append_kcore_note(notes, &i, CORE_STR, NT_TASKSTRUCT, current,
				  arch_task_struct_size);
		/*
		 * vmcoreinfo_size is mostly constant after init time, but it
		 * can be changed by crash_save_vmcoreinfo(). Racing here with a
		 * panic on another CPU before the machine goes down is insanely
		 * unlikely, but it's better to not leave potential buffer
		 * overflows lying around, regardless.
		 */
		append_kcore_note(notes, &i, VMCOREINFO_NOTE_NAME, 0,
				  vmcoreinfo_data,
				  min(vmcoreinfo_size, notes_len - i));

		tsz = min_t(size_t, buflen, notes_offset + notes_len - *fpos);
		if (copy_to_user(buffer, notes + *fpos - notes_offset, tsz)) {
			kfree(notes);
			ret = -EFAULT;
			goto out;
		}
		kfree(notes);

		buffer += tsz;
		buflen -= tsz;
		*fpos += tsz;
	}

	/*
	 * Check to see if our file offset matches with any of
	 * the addresses in the elf_phdr on our list.
	 */
	start = kc_offset_to_vaddr(*fpos - data_offset);
	if ((tsz = (PAGE_SIZE - (start & ~PAGE_MASK))) > buflen)
		tsz = buflen;

	m = NULL;
	while (buflen) {
		/*
		 * If this is the first iteration or the address is not within
		 * the previous entry, search for a matching entry.
		 */
		if (!m || start < m->addr || start >= m->addr + m->size) {
			list_for_each_entry(m, &kclist_head, list) {
				if (start >= m->addr &&
				    start < m->addr + m->size)
					break;
			}
		}

		if (&m->list == &kclist_head) {
			if (clear_user(buffer, tsz)) {
				ret = -EFAULT;
				goto out;
			}
			m = NULL;	/* skip the list anchor */
		} else if (m->type == KCORE_VMALLOC) {
			vread(buf, (char *)start, tsz);
			/* we have to zero-fill user buffer even if no read */
			if (copy_to_user(buffer, buf, tsz)) {
				ret = -EFAULT;
				goto out;
			}
		} else if (m->type == KCORE_USER) {
			/* User page is handled prior to normal kernel page: */
			if (copy_to_user(buffer, (char *)start, tsz)) {
				ret = -EFAULT;
				goto out;
			}
		} else {
			if (kern_addr_valid(start)) {
				/*
				 * Using bounce buffer to bypass the
				 * hardened user copy kernel text checks.
				 */
				if (probe_kernel_read(buf, (void *) start, tsz)) {
					if (clear_user(buffer, tsz)) {
						ret = -EFAULT;
						goto out;
					}
				} else {
					if (copy_to_user(buffer, buf, tsz)) {
						ret = -EFAULT;
						goto out;
					}
				}
			} else {
				if (clear_user(buffer, tsz)) {
					ret = -EFAULT;
					goto out;
				}
			}
		}
		buflen -= tsz;
		*fpos += tsz;
		buffer += tsz;
		start += tsz;
		tsz = (buflen > PAGE_SIZE ? PAGE_SIZE : buflen);
	}

out:
	up_read(&kclist_lock);
	if (ret)
		return ret;
	return orig_buflen - buflen;
}

static int open_kcore(struct inode *inode, struct file *filp)
{
	if (!capable(CAP_SYS_RAWIO))
		return -EPERM;

	filp->private_data = kmalloc(PAGE_SIZE, GFP_KERNEL);
	if (!filp->private_data)
		return -ENOMEM;

	if (kcore_need_update)
		kcore_update_ram();
	if (i_size_read(inode) != proc_root_kcore->size) {
		inode_lock(inode);
		i_size_write(inode, proc_root_kcore->size);
		inode_unlock(inode);
	}
	return 0;
}

static int release_kcore(struct inode *inode, struct file *file)
{
	kfree(file->private_data);
	return 0;
}

static const struct file_operations proc_kcore_operations = {
	.read		= read_kcore,
	.open		= open_kcore,
	.release	= release_kcore,
	.llseek		= default_llseek,
};

/* just remember that we have to update kcore */
static int __meminit kcore_callback(struct notifier_block *self,
				    unsigned long action, void *arg)
{
	switch (action) {
	case MEM_ONLINE:
	case MEM_OFFLINE:
		kcore_need_update = 1;
		break;
	}
	return NOTIFY_OK;
}

static struct notifier_block kcore_callback_nb __meminitdata = {
	.notifier_call = kcore_callback,
	.priority = 0,
};

static struct kcore_list kcore_vmalloc;

#ifdef CONFIG_ARCH_PROC_KCORE_TEXT
static struct kcore_list kcore_text;
/*
 * If defined, special segment is used for mapping kernel text instead of
 * direct-map area. We need to create special TEXT section.
 */
static void __init proc_kcore_text_init(void)
{
	kclist_add(&kcore_text, _text, _end - _text, KCORE_TEXT);
}
#else
static void __init proc_kcore_text_init(void)
{
}
#endif

#if defined(CONFIG_MODULES) && defined(MODULES_VADDR)
/*
 * MODULES_VADDR has no intersection with VMALLOC_ADDR.
 */
struct kcore_list kcore_modules;
static void __init add_modules_range(void)
{
	if (MODULES_VADDR != VMALLOC_START && MODULES_END != VMALLOC_END) {
		kclist_add(&kcore_modules, (void *)MODULES_VADDR,
			MODULES_END - MODULES_VADDR, KCORE_VMALLOC);
	}
}
#else
static void __init add_modules_range(void)
{
}
#endif

static int __init proc_kcore_init(void)
{
	proc_root_kcore = proc_create("kcore", S_IRUSR, NULL,
				      &proc_kcore_operations);
	if (!proc_root_kcore) {
		pr_err("couldn't create /proc/kcore\n");
		return 0; /* Always returns 0. */
	}
	/* Store text area if it's special */
	proc_kcore_text_init();
	/* Store vmalloc area */
	kclist_add(&kcore_vmalloc, (void *)VMALLOC_START,
		VMALLOC_END - VMALLOC_START, KCORE_VMALLOC);
	add_modules_range();
	/* Store direct-map area from physical memory map */
	kcore_update_ram();
	register_hotmemory_notifier(&kcore_callback_nb);

	return 0;
}
fs_initcall(proc_kcore_init);
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
1da177e4 LT	2	/*
	3	* fs/proc/kcore.c kernel ELF core dumper
	4	*
	5	* Modelled on fs/exec.c:aout_core_dump()
	6	* Jeremy Fitzhardinge <[email protected]>
	7	* ELF version written by David Howells <[email protected]>
	8	* Modified and incorporated into 2.3.x by Tigran Aivazian <[email protected]>
	9	* Support to dump vmalloc'd areas (ELF only), Tigran Aivazian <[email protected]>
	10	* Safe accesses to vmalloc/direct-mapped discontiguous areas, Kanoj Sarcar <[email protected]>
	11	*/
	12
23c85094	13	#include <linux/crash_core.h>
1da177e4 LT	14	#include <linux/mm.h>
1da177e4 LT	15	#include <linux/proc_fs.h>
2f96b8c1	16	#include <linux/kcore.h>
1da177e4	17	#include <linux/user.h>
16f7e0fe	18	#include <linux/capability.h>
1da177e4 LT	19	#include <linux/elf.h>
1da177e4 LT	20	#include <linux/elfcore.h>
3c743a7f	21	#include <linux/notifier.h>
1da177e4 LT	22	#include <linux/vmalloc.h>
1da177e4 LT	23	#include <linux/highmem.h>
87ebdc00	24	#include <linux/printk.h>
3089aa1b	25	#include <linux/bootmem.h>
1da177e4	26	#include <linux/init.h>
5a0e3ad6	27	#include <linux/slab.h>
7c0f6ba6	28	#include <linux/uaccess.h>
1da177e4	29	#include <asm/io.h>
2ef43ec7	30	#include <linux/list.h>
3089aa1b	31	#include <linux/ioport.h>
3089aa1b	32	#include <linux/memory.h>
29930025	33	#include <linux/sched/task.h>
9492587c	34	#include <asm/sections.h>
59d8053f	35	#include "internal.h"
1da177e4	36
36027604	37	#define CORE_STR "CORE"
1da177e4	38
79885b22 EI	39	#ifndef ELF_CORE_EFLAGS
	40	#define ELF_CORE_EFLAGS 0
	41	#endif
	42
97ce5d6d AD	43	static struct proc_dir_entry *proc_root_kcore;
97ce5d6d AD	44
1da177e4 LT	45
	46	#ifndef kc_vaddr_to_offset
	47	#define kc_vaddr_to_offset(v) ((v) - PAGE_OFFSET)
	48	#endif
	49	#ifndef kc_offset_to_vaddr
	50	#define kc_offset_to_vaddr(o) ((o) + PAGE_OFFSET)
	51	#endif
	52
2ef43ec7	53	static LIST_HEAD(kclist_head);
0b172f84	54	static DECLARE_RWSEM(kclist_lock);
3089aa1b	55	static int kcore_need_update = 1;
1da177e4	56
a8dd9c4d OS	57	/* This doesn't grab kclist_lock, so it should only be used at init time. */
	58	void __init kclist_add(struct kcore_list new, void addr, size_t size,
	59	int type)
1da177e4 LT	60	{
	61	new->addr = (unsigned long)addr;
	62	new->size = size;
c30bb2a2	63	new->type = type;
1da177e4	64
2ef43ec7	65	list_add_tail(&new->list, &kclist_head);
1da177e4 LT	66	}
1da177e4 LT	67
37e949bd OS	68	static size_t get_kcore_size(int nphdr, size_t phdrs_len, size_t *notes_len,
37e949bd OS	69	size_t *data_offset)
1da177e4 LT	70	{
	71	size_t try, size;
	72	struct kcore_list *m;
	73
	74	nphdr = 1; / PT_NOTE */
	75	size = 0;
	76
2ef43ec7	77	list_for_each_entry(m, &kclist_head, list) {
1da177e4 LT	78	try = kc_vaddr_to_offset((size_t)m->addr + m->size);
	79	if (try > size)
	80	size = try;
	81	nphdr = nphdr + 1;
	82	}
37e949bd OS	83
37e949bd OS	84	phdrs_len = nphdr * sizeof(struct elf_phdr);
23c85094 OS	85	notes_len = (4 sizeof(struct elf_note) +
	86	3 * ALIGN(sizeof(CORE_STR), 4) +
	87	VMCOREINFO_NOTE_NAME_BYTES +
37e949bd OS	88	ALIGN(sizeof(struct elf_prstatus), 4) +
37e949bd OS	89	ALIGN(sizeof(struct elf_prpsinfo), 4) +
23c85094 OS	90	ALIGN(arch_task_struct_size, 4) +
23c85094 OS	91	ALIGN(vmcoreinfo_size, 4));
37e949bd OS	92	data_offset = PAGE_ALIGN(sizeof(struct elfhdr) + phdrs_len +
	93	*notes_len);
	94	return *data_offset + size;
1da177e4 LT	95	}
1da177e4 LT	96
3089aa1b KH	97	#ifdef CONFIG_HIGHMEM
	98	/*
	99	* If no highmem, we can assume [0...max_low_pfn) continuous range of memory
	100	* because memory hole is not as big as !HIGHMEM case.
	101	* (HIGHMEM is special because part of memory is _invisible_ from the kernel.)
	102	*/
b66fb005	103	static int kcore_ram_list(struct list_head *head)
3089aa1b	104	{
3089aa1b	105	struct kcore_list *ent;
3089aa1b KH	106
	107	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
	108	if (!ent)
	109	return -ENOMEM;
	110	ent->addr = (unsigned long)__va(0);
	111	ent->size = max_low_pfn << PAGE_SHIFT;
	112	ent->type = KCORE_RAM;
b66fb005 OS	113	list_add(&ent->list, head);
b66fb005 OS	114	return 0;
3089aa1b KH	115	}
	116
	117	#else /* !CONFIG_HIGHMEM */
	118
26562c59 KH	119	#ifdef CONFIG_SPARSEMEM_VMEMMAP
26562c59 KH	120	/* calculate vmemmap's address from given system ram pfn and register it */
b908243c DH	121	static int
b908243c DH	122	get_sparsemem_vmemmap_info(struct kcore_list ent, struct list_head head)
26562c59 KH	123	{
	124	unsigned long pfn = __pa(ent->addr) >> PAGE_SHIFT;
	125	unsigned long nr_pages = ent->size >> PAGE_SHIFT;
	126	unsigned long start, end;
	127	struct kcore_list vmm, tmp;
	128
	129
	130	start = ((unsigned long)pfn_to_page(pfn)) & PAGE_MASK;
	131	end = ((unsigned long)pfn_to_page(pfn + nr_pages)) - 1;
108a8a11	132	end = PAGE_ALIGN(end);
26562c59 KH	133	/* overlap check (because we have to align page */
	134	list_for_each_entry(tmp, head, list) {
	135	if (tmp->type != KCORE_VMEMMAP)
	136	continue;
	137	if (start < tmp->addr + tmp->size)
	138	if (end > tmp->addr)
	139	end = tmp->addr;
	140	}
	141	if (start < end) {
	142	vmm = kmalloc(sizeof(*vmm), GFP_KERNEL);
	143	if (!vmm)
	144	return 0;
	145	vmm->addr = start;
	146	vmm->size = end - start;
	147	vmm->type = KCORE_VMEMMAP;
	148	list_add_tail(&vmm->list, head);
	149	}
	150	return 1;
	151
	152	}
	153	#else
b908243c DH	154	static int
b908243c DH	155	get_sparsemem_vmemmap_info(struct kcore_list ent, struct list_head head)
26562c59 KH	156	{
	157	return 1;
	158	}
	159
	160	#endif
	161
3089aa1b KH	162	static int
	163	kclist_add_private(unsigned long pfn, unsigned long nr_pages, void *arg)
	164	{
	165	struct list_head head = (struct list_head )arg;
	166	struct kcore_list *ent;
3955333d LA	167	struct page *p;
	168
	169	if (!pfn_valid(pfn))
	170	return 1;
	171
	172	p = pfn_to_page(pfn);
	173	if (!memmap_valid_within(pfn, p, page_zone(p)))
	174	return 1;
3089aa1b KH	175
	176	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
	177	if (!ent)
	178	return -ENOMEM;
3955333d	179	ent->addr = (unsigned long)page_to_virt(p);
3089aa1b KH	180	ent->size = nr_pages << PAGE_SHIFT;
3089aa1b KH	181
3955333d	182	if (!virt_addr_valid(ent->addr))
3089aa1b KH	183	goto free_out;
	184
	185	/* cut not-mapped area. ....from ppc-32 code. */
	186	if (ULONG_MAX - ent->addr < ent->size)
	187	ent->size = ULONG_MAX - ent->addr;
	188
3955333d LA	189	/*
	190	* We've already checked virt_addr_valid so we know this address
	191	* is a valid pointer, therefore we can check against it to determine
	192	* if we need to trim
	193	*/
	194	if (VMALLOC_START > ent->addr) {
3089aa1b KH	195	if (VMALLOC_START - ent->addr < ent->size)
	196	ent->size = VMALLOC_START - ent->addr;
	197	}
	198
	199	ent->type = KCORE_RAM;
	200	list_add_tail(&ent->list, head);
26562c59 KH	201
	202	if (!get_sparsemem_vmemmap_info(ent, head)) {
	203	list_del(&ent->list);
	204	goto free_out;
	205	}
	206
3089aa1b KH	207	return 0;
	208	free_out:
	209	kfree(ent);
	210	return 1;
	211	}
	212
b66fb005	213	static int kcore_ram_list(struct list_head *list)
3089aa1b KH	214	{
	215	int nid, ret;
	216	unsigned long end_pfn;
3089aa1b KH	217
	218	/* Not inialized....update now */
	219	/* find out "max pfn" */
	220	end_pfn = 0;
4ff1b2c2	221	for_each_node_state(nid, N_MEMORY) {
3089aa1b	222	unsigned long node_end;
83285c72	223	node_end = node_end_pfn(nid);
3089aa1b KH	224	if (end_pfn < node_end)
	225	end_pfn = node_end;
	226	}
	227	/* scan 0 to max_pfn */
b66fb005 OS	228	ret = walk_system_ram_range(0, end_pfn, list, kclist_add_private);
b66fb005 OS	229	if (ret)
3089aa1b	230	return -ENOMEM;
b66fb005 OS	231	return 0;
	232	}
	233	#endif /* CONFIG_HIGHMEM */
	234
	235	static int kcore_update_ram(void)
	236	{
	237	LIST_HEAD(list);
	238	LIST_HEAD(garbage);
	239	int nphdr;
37e949bd	240	size_t phdrs_len, notes_len, data_offset;
b66fb005 OS	241	struct kcore_list tmp, pos;
	242	int ret = 0;
	243
	244	down_write(&kclist_lock);
	245	if (!xchg(&kcore_need_update, 0))
	246	goto out;
	247
	248	ret = kcore_ram_list(&list);
	249	if (ret) {
	250	/* Couldn't get the RAM list, try again next time. */
	251	WRITE_ONCE(kcore_need_update, 1);
	252	list_splice_tail(&list, &garbage);
	253	goto out;
	254	}
	255
	256	list_for_each_entry_safe(pos, tmp, &kclist_head, list) {
	257	if (pos->type == KCORE_RAM \|\| pos->type == KCORE_VMEMMAP)
	258	list_move(&pos->list, &garbage);
	259	}
	260	list_splice_tail(&list, &kclist_head);
	261
37e949bd OS	262	proc_root_kcore->size = get_kcore_size(&nphdr, &phdrs_len, &notes_len,
37e949bd OS	263	&data_offset);
b66fb005 OS	264
	265	out:
	266	up_write(&kclist_lock);
	267	list_for_each_entry_safe(pos, tmp, &garbage, list) {
	268	list_del(&pos->list);
	269	kfree(pos);
3089aa1b	270	}
3089aa1b KH	271	return ret;
3089aa1b KH	272	}
1da177e4	273
37e949bd OS	274	static void append_kcore_note(char notes, size_t i, const char *name,
	275	unsigned int type, const void *desc,
	276	size_t descsz)
1da177e4	277	{
37e949bd OS	278	struct elf_note note = (struct elf_note )&notes[*i];
	279
	280	note->n_namesz = strlen(name) + 1;
	281	note->n_descsz = descsz;
	282	note->n_type = type;
	283	i += sizeof(note);
	284	memcpy(&notes[*i], name, note->n_namesz);
	285	i = ALIGN(i + note->n_namesz, 4);
	286	memcpy(&notes[*i], desc, descsz);
	287	i = ALIGN(i + descsz, 4);
	288	}
1da177e4	289
1da177e4 LT	290	static ssize_t
	291	read_kcore(struct file file, char __user buffer, size_t buflen, loff_t *fpos)
	292	{
f5beeb18	293	char *buf = file->private_data;
37e949bd OS	294	size_t phdrs_offset, notes_offset, data_offset;
	295	size_t phdrs_len, notes_len;
	296	struct kcore_list *m;
	297	size_t tsz;
1da177e4 LT	298	int nphdr;
1da177e4 LT	299	unsigned long start;
3673fb08 OS	300	size_t orig_buflen = buflen;
3673fb08 OS	301	int ret = 0;
1da177e4	302
0b172f84	303	down_read(&kclist_lock);
678ad5d8	304
37e949bd OS	305	get_kcore_size(&nphdr, &phdrs_len, &notes_len, &data_offset);
	306	phdrs_offset = sizeof(struct elfhdr);
	307	notes_offset = phdrs_offset + phdrs_len;
	308
	309	/* ELF file header. */
	310	if (buflen && *fpos < sizeof(struct elfhdr)) {
	311	struct elfhdr ehdr = {
	312	.e_ident = {
	313	[EI_MAG0] = ELFMAG0,
	314	[EI_MAG1] = ELFMAG1,
	315	[EI_MAG2] = ELFMAG2,
	316	[EI_MAG3] = ELFMAG3,
	317	[EI_CLASS] = ELF_CLASS,
	318	[EI_DATA] = ELF_DATA,
	319	[EI_VERSION] = EV_CURRENT,
	320	[EI_OSABI] = ELF_OSABI,
	321	},
	322	.e_type = ET_CORE,
	323	.e_machine = ELF_ARCH,
	324	.e_version = EV_CURRENT,
	325	.e_phoff = sizeof(struct elfhdr),
	326	.e_flags = ELF_CORE_EFLAGS,
	327	.e_ehsize = sizeof(struct elfhdr),
	328	.e_phentsize = sizeof(struct elf_phdr),
	329	.e_phnum = nphdr,
	330	};
	331
	332	tsz = min_t(size_t, buflen, sizeof(struct elfhdr) - *fpos);
	333	if (copy_to_user(buffer, (char )&ehdr + fpos, tsz)) {
	334	ret = -EFAULT;
	335	goto out;
	336	}
1da177e4	337
37e949bd OS	338	buffer += tsz;
	339	buflen -= tsz;
	340	*fpos += tsz;
	341	}
1da177e4	342
37e949bd OS	343	/* ELF program headers. */
	344	if (buflen && *fpos < phdrs_offset + phdrs_len) {
	345	struct elf_phdr phdrs, phdr;
1da177e4	346
37e949bd OS	347	phdrs = kzalloc(phdrs_len, GFP_KERNEL);
37e949bd OS	348	if (!phdrs) {
3673fb08 OS	349	ret = -ENOMEM;
3673fb08 OS	350	goto out;
1da177e4	351	}
37e949bd OS	352
	353	phdrs[0].p_type = PT_NOTE;
	354	phdrs[0].p_offset = notes_offset;
	355	phdrs[0].p_filesz = notes_len;
	356
	357	phdr = &phdrs[1];
	358	list_for_each_entry(m, &kclist_head, list) {
	359	phdr->p_type = PT_LOAD;
	360	phdr->p_flags = PF_R \| PF_W \| PF_X;
	361	phdr->p_offset = kc_vaddr_to_offset(m->addr) + data_offset;
d207ea8e LT	362	if (m->type == KCORE_REMAP)
	363	phdr->p_vaddr = (size_t)m->vaddr;
	364	else
	365	phdr->p_vaddr = (size_t)m->addr;
	366	if (m->type == KCORE_RAM \|\| m->type == KCORE_REMAP)
37e949bd OS	367	phdr->p_paddr = __pa(m->addr);
	368	else if (m->type == KCORE_TEXT)
	369	phdr->p_paddr = __pa_symbol(m->addr);
	370	else
	371	phdr->p_paddr = (elf_addr_t)-1;
	372	phdr->p_filesz = phdr->p_memsz = m->size;
	373	phdr->p_align = PAGE_SIZE;
	374	phdr++;
	375	}
	376
	377	tsz = min_t(size_t, buflen, phdrs_offset + phdrs_len - *fpos);
	378	if (copy_to_user(buffer, (char )phdrs + fpos - phdrs_offset,
	379	tsz)) {
	380	kfree(phdrs);
3673fb08 OS	381	ret = -EFAULT;
3673fb08 OS	382	goto out;
1da177e4	383	}
37e949bd OS	384	kfree(phdrs);
	385
	386	buffer += tsz;
1da177e4 LT	387	buflen -= tsz;
1da177e4 LT	388	*fpos += tsz;
37e949bd OS	389	}
	390
	391	/* ELF note segment. */
	392	if (buflen && *fpos < notes_offset + notes_len) {
	393	struct elf_prstatus prstatus = {};
	394	struct elf_prpsinfo prpsinfo = {
	395	.pr_sname = 'R',
	396	.pr_fname = "vmlinux",
	397	};
	398	char *notes;
	399	size_t i = 0;
	400
	401	strlcpy(prpsinfo.pr_psargs, saved_command_line,
	402	sizeof(prpsinfo.pr_psargs));
	403
	404	notes = kzalloc(notes_len, GFP_KERNEL);
	405	if (!notes) {
	406	ret = -ENOMEM;
	407	goto out;
	408	}
	409
	410	append_kcore_note(notes, &i, CORE_STR, NT_PRSTATUS, &prstatus,
	411	sizeof(prstatus));
	412	append_kcore_note(notes, &i, CORE_STR, NT_PRPSINFO, &prpsinfo,
	413	sizeof(prpsinfo));
	414	append_kcore_note(notes, &i, CORE_STR, NT_TASKSTRUCT, current,
	415	arch_task_struct_size);
23c85094 OS	416	/*
	417	* vmcoreinfo_size is mostly constant after init time, but it
	418	* can be changed by crash_save_vmcoreinfo(). Racing here with a
	419	* panic on another CPU before the machine goes down is insanely
	420	* unlikely, but it's better to not leave potential buffer
	421	* overflows lying around, regardless.
	422	*/
	423	append_kcore_note(notes, &i, VMCOREINFO_NOTE_NAME, 0,
	424	vmcoreinfo_data,
	425	min(vmcoreinfo_size, notes_len - i));
1da177e4	426
37e949bd OS	427	tsz = min_t(size_t, buflen, notes_offset + notes_len - *fpos);
	428	if (copy_to_user(buffer, notes + *fpos - notes_offset, tsz)) {
	429	kfree(notes);
	430	ret = -EFAULT;
3673fb08	431	goto out;
37e949bd OS	432	}
	433	kfree(notes);
	434
	435	buffer += tsz;
	436	buflen -= tsz;
	437	*fpos += tsz;
3673fb08	438	}
1da177e4 LT	439
	440	/*
	441	* Check to see if our file offset matches with any of
	442	* the addresses in the elf_phdr on our list.
	443	*/
37e949bd	444	start = kc_offset_to_vaddr(*fpos - data_offset);
1da177e4 LT	445	if ((tsz = (PAGE_SIZE - (start & ~PAGE_MASK))) > buflen)
1da177e4 LT	446	tsz = buflen;
1da177e4	447
bf991c22	448	m = NULL;
37e949bd	449	while (buflen) {
bf991c22 OS	450	/*
	451	* If this is the first iteration or the address is not within
	452	* the previous entry, search for a matching entry.
	453	*/
	454	if (!m \|\| start < m->addr \|\| start >= m->addr + m->size) {
	455	list_for_each_entry(m, &kclist_head, list) {
	456	if (start >= m->addr &&
	457	start < m->addr + m->size)
	458	break;
	459	}
1da177e4	460	}
1da177e4	461
4fd2c20d	462	if (&m->list == &kclist_head) {
3673fb08 OS	463	if (clear_user(buffer, tsz)) {
	464	ret = -EFAULT;
	465	goto out;
	466	}
a1b3d2f2	467	m = NULL; /* skip the list anchor */
737326aa	468	} else if (m->type == KCORE_VMALLOC) {
f5beeb18	469	vread(buf, (char *)start, tsz);
73d7c33e	470	/* we have to zero-fill user buffer even if no read */
3673fb08 OS	471	if (copy_to_user(buffer, buf, tsz)) {
	472	ret = -EFAULT;
	473	goto out;
	474	}
595dd46e JZ	475	} else if (m->type == KCORE_USER) {
595dd46e JZ	476	/* User page is handled prior to normal kernel page: */
3673fb08 OS	477	if (copy_to_user(buffer, (char *)start, tsz)) {
	478	ret = -EFAULT;
	479	goto out;
	480	}
1da177e4 LT	481	} else {
1da177e4 LT	482	if (kern_addr_valid(start)) {
df04abfd JO	483	/*
	484	* Using bounce buffer to bypass the
	485	* hardened user copy kernel text checks.
	486	*/
d0290bc2	487	if (probe_kernel_read(buf, (void *) start, tsz)) {
3673fb08 OS	488	if (clear_user(buffer, tsz)) {
	489	ret = -EFAULT;
	490	goto out;
	491	}
d0290bc2	492	} else {
3673fb08 OS	493	if (copy_to_user(buffer, buf, tsz)) {
	494	ret = -EFAULT;
	495	goto out;
	496	}
1da177e4 LT	497	}
1da177e4 LT	498	} else {
3673fb08 OS	499	if (clear_user(buffer, tsz)) {
	500	ret = -EFAULT;
	501	goto out;
	502	}
1da177e4 LT	503	}
	504	}
	505	buflen -= tsz;
	506	*fpos += tsz;
	507	buffer += tsz;
1da177e4 LT	508	start += tsz;
	509	tsz = (buflen > PAGE_SIZE ? PAGE_SIZE : buflen);
	510	}
	511
3673fb08 OS	512	out:
	513	up_read(&kclist_lock);
	514	if (ret)
	515	return ret;
	516	return orig_buflen - buflen;
1da177e4	517	}
97ce5d6d	518
3089aa1b KH	519	static int open_kcore(struct inode inode, struct file filp)
	520	{
	521	if (!capable(CAP_SYS_RAWIO))
	522	return -EPERM;
f5beeb18 JO	523
	524	filp->private_data = kmalloc(PAGE_SIZE, GFP_KERNEL);
	525	if (!filp->private_data)
	526	return -ENOMEM;
	527
3089aa1b KH	528	if (kcore_need_update)
3089aa1b KH	529	kcore_update_ram();
0d4c36a9	530	if (i_size_read(inode) != proc_root_kcore->size) {
5955102c	531	inode_lock(inode);
0d4c36a9	532	i_size_write(inode, proc_root_kcore->size);
5955102c	533	inode_unlock(inode);
0d4c36a9	534	}
3089aa1b KH	535	return 0;
	536	}
	537
f5beeb18 JO	538	static int release_kcore(struct inode inode, struct file file)
	539	{
	540	kfree(file->private_data);
	541	return 0;
	542	}
3089aa1b KH	543
	544	static const struct file_operations proc_kcore_operations = {
	545	.read = read_kcore,
	546	.open = open_kcore,
f5beeb18	547	.release = release_kcore,
ceff1a77	548	.llseek = default_llseek,
3089aa1b KH	549	};
3089aa1b KH	550
3089aa1b KH	551	/* just remember that we have to update kcore */
	552	static int __meminit kcore_callback(struct notifier_block *self,
	553	unsigned long action, void *arg)
	554	{
	555	switch (action) {
	556	case MEM_ONLINE:
	557	case MEM_OFFLINE:
3089aa1b	558	kcore_need_update = 1;
bf531831	559	break;
3089aa1b KH	560	}
	561	return NOTIFY_OK;
	562	}
3089aa1b	563
3c743a7f AM	564	static struct notifier_block kcore_callback_nb __meminitdata = {
	565	.notifier_call = kcore_callback,
	566	.priority = 0,
	567	};
3089aa1b	568
a0614da8 KH	569	static struct kcore_list kcore_vmalloc;
a0614da8 KH	570
9492587c KH	571	#ifdef CONFIG_ARCH_PROC_KCORE_TEXT
	572	static struct kcore_list kcore_text;
	573	/*
	574	* If defined, special segment is used for mapping kernel text instead of
	575	* direct-map area. We need to create special TEXT section.
	576	*/
	577	static void __init proc_kcore_text_init(void)
	578	{
36e15263	579	kclist_add(&kcore_text, _text, _end - _text, KCORE_TEXT);
9492587c KH	580	}
	581	#else
	582	static void __init proc_kcore_text_init(void)
	583	{
	584	}
	585	#endif
	586
81ac3ad9 KH	587	#if defined(CONFIG_MODULES) && defined(MODULES_VADDR)
	588	/*
	589	* MODULES_VADDR has no intersection with VMALLOC_ADDR.
	590	*/
	591	struct kcore_list kcore_modules;
	592	static void __init add_modules_range(void)
	593	{
bf3e2692 BH	594	if (MODULES_VADDR != VMALLOC_START && MODULES_END != VMALLOC_END) {
bf3e2692 BH	595	kclist_add(&kcore_modules, (void *)MODULES_VADDR,
81ac3ad9	596	MODULES_END - MODULES_VADDR, KCORE_VMALLOC);
bf3e2692	597	}
81ac3ad9 KH	598	}
	599	#else
	600	static void __init add_modules_range(void)
	601	{
	602	}
	603	#endif
	604
97ce5d6d AD	605	static int __init proc_kcore_init(void)
97ce5d6d AD	606	{
3089aa1b KH	607	proc_root_kcore = proc_create("kcore", S_IRUSR, NULL,
3089aa1b KH	608	&proc_kcore_operations);
90396f96	609	if (!proc_root_kcore) {
87ebdc00	610	pr_err("couldn't create /proc/kcore\n");
90396f96 KH	611	return 0; /* Always returns 0. */
90396f96 KH	612	}
3089aa1b	613	/* Store text area if it's special */
9492587c	614	proc_kcore_text_init();
3089aa1b	615	/* Store vmalloc area */
a0614da8 KH	616	kclist_add(&kcore_vmalloc, (void *)VMALLOC_START,
a0614da8 KH	617	VMALLOC_END - VMALLOC_START, KCORE_VMALLOC);
81ac3ad9	618	add_modules_range();
3089aa1b KH	619	/* Store direct-map area from physical memory map */
3089aa1b KH	620	kcore_update_ram();
3c743a7f	621	register_hotmemory_notifier(&kcore_callback_nb);
3089aa1b	622
97ce5d6d AD	623	return 0;
97ce5d6d AD	624	}
abaf3787	625	fs_initcall(proc_kcore_init);