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

/*
 *	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/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 <asm/uaccess.h>
#include <asm/io.h>
#include <linux/list.h>
#include <linux/ioport.h>
#include <linux/memory.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

/* An ELF note in memory */
struct memelfnote
{
	const char *name;
	int type;
	unsigned int datasz;
	void *data;
};

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

void
kclist_add(struct kcore_list *new, void *addr, size_t size, int type)
{
	new->addr = (unsigned long)addr;
	new->size = size;
	new->type = type;

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

static size_t get_kcore_size(int *nphdr, size_t *elf_buflen)
{
	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;
	}
	*elf_buflen =	sizeof(struct elfhdr) + 
			(*nphdr + 2)*sizeof(struct elf_phdr) + 
			3 * ((sizeof(struct elf_note)) +
			     roundup(sizeof(CORE_STR), 4)) +
			roundup(sizeof(struct elf_prstatus), 4) +
			roundup(sizeof(struct elf_prpsinfo), 4) +
			roundup(arch_task_struct_size, 4);
	*elf_buflen = PAGE_ALIGN(*elf_buflen);
	return size + *elf_buflen;
}

static void free_kclist_ents(struct list_head *head)
{
	struct kcore_list *tmp, *pos;

	list_for_each_entry_safe(pos, tmp, head, list) {
		list_del(&pos->list);
		kfree(pos);
	}
}
/*
 * Replace all KCORE_RAM/KCORE_VMEMMAP information with passed list.
 */
static void __kcore_update_ram(struct list_head *list)
{
	int nphdr;
	size_t size;
	struct kcore_list *tmp, *pos;
	LIST_HEAD(garbage);

	write_lock(&kclist_lock);
	if (kcore_need_update) {
		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);
	} else
		list_splice(list, &garbage);
	kcore_need_update = 0;
	proc_root_kcore->size = get_kcore_size(&nphdr, &size);
	write_unlock(&kclist_lock);

	free_kclist_ents(&garbage);
}


#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_update_ram(void)
{
	LIST_HEAD(head);
	struct kcore_list *ent;
	int ret = 0;

	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);
	__kcore_update_ram(&head);
	return ret;
}

#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;

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

	/* Sanity check: Can happen in 32bit arch...maybe */
	if (ent->addr < (unsigned long) __va(0))
		goto free_out;

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

	/* cut when vmalloc() area is higher than direct-map area */
	if (VMALLOC_START > (unsigned long)__va(0)) {
		if (ent->addr > VMALLOC_START)
			goto free_out;
		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_update_ram(void)
{
	int nid, ret;
	unsigned long end_pfn;
	LIST_HEAD(head);

	/* 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, &head, kclist_add_private);
	if (ret) {
		free_kclist_ents(&head);
		return -ENOMEM;
	}
	__kcore_update_ram(&head);
	return ret;
}
#endif /* CONFIG_HIGHMEM */

/*****************************************************************************/
/*
 * determine size of ELF note
 */
static int notesize(struct memelfnote *en)
{
	int sz;

	sz = sizeof(struct elf_note);
	sz += roundup((strlen(en->name) + 1), 4);
	sz += roundup(en->datasz, 4);

	return sz;
} /* end notesize() */

/*****************************************************************************/
/*
 * store a note in the header buffer
 */
static char *storenote(struct memelfnote *men, char *bufp)
{
	struct elf_note en;

#define DUMP_WRITE(addr,nr) do { memcpy(bufp,addr,nr); bufp += nr; } while(0)

	en.n_namesz = strlen(men->name) + 1;
	en.n_descsz = men->datasz;
	en.n_type = men->type;

	DUMP_WRITE(&en, sizeof(en));
	DUMP_WRITE(men->name, en.n_namesz);

	/* XXX - cast from long long to long to avoid need for libgcc.a */
	bufp = (char*) roundup((unsigned long)bufp,4);
	DUMP_WRITE(men->data, men->datasz);
	bufp = (char*) roundup((unsigned long)bufp,4);

#undef DUMP_WRITE

	return bufp;
} /* end storenote() */

/*
 * store an ELF coredump header in the supplied buffer
 * nphdr is the number of elf_phdr to insert
 */
static void elf_kcore_store_hdr(char *bufp, int nphdr, int dataoff)
{
	struct elf_prstatus prstatus;	/* NT_PRSTATUS */
	struct elf_prpsinfo prpsinfo;	/* NT_PRPSINFO */
	struct elf_phdr *nhdr, *phdr;
	struct elfhdr *elf;
	struct memelfnote notes[3];
	off_t offset = 0;
	struct kcore_list *m;

	/* setup ELF header */
	elf = (struct elfhdr *) bufp;
	bufp += sizeof(struct elfhdr);
	offset += sizeof(struct elfhdr);
	memcpy(elf->e_ident, ELFMAG, SELFMAG);
	elf->e_ident[EI_CLASS]	= ELF_CLASS;
	elf->e_ident[EI_DATA]	= ELF_DATA;
	elf->e_ident[EI_VERSION]= EV_CURRENT;
	elf->e_ident[EI_OSABI] = ELF_OSABI;
	memset(elf->e_ident+EI_PAD, 0, EI_NIDENT-EI_PAD);
	elf->e_type	= ET_CORE;
	elf->e_machine	= ELF_ARCH;
	elf->e_version	= EV_CURRENT;
	elf->e_entry	= 0;
	elf->e_phoff	= sizeof(struct elfhdr);
	elf->e_shoff	= 0;
	elf->e_flags	= ELF_CORE_EFLAGS;
	elf->e_ehsize	= sizeof(struct elfhdr);
	elf->e_phentsize= sizeof(struct elf_phdr);
	elf->e_phnum	= nphdr;
	elf->e_shentsize= 0;
	elf->e_shnum	= 0;
	elf->e_shstrndx	= 0;

	/* setup ELF PT_NOTE program header */
	nhdr = (struct elf_phdr *) bufp;
	bufp += sizeof(struct elf_phdr);
	offset += sizeof(struct elf_phdr);
	nhdr->p_type	= PT_NOTE;
	nhdr->p_offset	= 0;
	nhdr->p_vaddr	= 0;
	nhdr->p_paddr	= 0;
	nhdr->p_filesz	= 0;
	nhdr->p_memsz	= 0;
	nhdr->p_flags	= 0;
	nhdr->p_align	= 0;

	/* setup ELF PT_LOAD program header for every area */
	list_for_each_entry(m, &kclist_head, list) {
		phdr = (struct elf_phdr *) bufp;
		bufp += sizeof(struct elf_phdr);
		offset += sizeof(struct elf_phdr);

		phdr->p_type	= PT_LOAD;
		phdr->p_flags	= PF_R|PF_W|PF_X;
		phdr->p_offset	= kc_vaddr_to_offset(m->addr) + dataoff;
		phdr->p_vaddr	= (size_t)m->addr;
		phdr->p_paddr	= 0;
		phdr->p_filesz	= phdr->p_memsz	= m->size;
		phdr->p_align	= PAGE_SIZE;
	}

	/*
	 * Set up the notes in similar form to SVR4 core dumps made
	 * with info from their /proc.
	 */
	nhdr->p_offset	= offset;

	/* set up the process status */
	notes[0].name = CORE_STR;
	notes[0].type = NT_PRSTATUS;
	notes[0].datasz = sizeof(struct elf_prstatus);
	notes[0].data = &prstatus;

	memset(&prstatus, 0, sizeof(struct elf_prstatus));

	nhdr->p_filesz	= notesize(&notes[0]);
	bufp = storenote(&notes[0], bufp);

	/* set up the process info */
	notes[1].name	= CORE_STR;
	notes[1].type	= NT_PRPSINFO;
	notes[1].datasz	= sizeof(struct elf_prpsinfo);
	notes[1].data	= &prpsinfo;

	memset(&prpsinfo, 0, sizeof(struct elf_prpsinfo));
	prpsinfo.pr_state	= 0;
	prpsinfo.pr_sname	= 'R';
	prpsinfo.pr_zomb	= 0;

	strcpy(prpsinfo.pr_fname, "vmlinux");
	strlcpy(prpsinfo.pr_psargs, saved_command_line, sizeof(prpsinfo.pr_psargs));

	nhdr->p_filesz	+= notesize(&notes[1]);
	bufp = storenote(&notes[1], bufp);

	/* set up the task structure */
	notes[2].name	= CORE_STR;
	notes[2].type	= NT_TASKSTRUCT;
	notes[2].datasz	= arch_task_struct_size;
	notes[2].data	= current;

	nhdr->p_filesz	+= notesize(&notes[2]);
	bufp = storenote(&notes[2], bufp);

} /* end elf_kcore_store_hdr() */

/*****************************************************************************/
/*
 * read from the ELF header and then kernel memory
 */
static ssize_t
read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)
{
	ssize_t acc = 0;
	size_t size, tsz;
	size_t elf_buflen;
	int nphdr;
	unsigned long start;

	read_lock(&kclist_lock);
	size = get_kcore_size(&nphdr, &elf_buflen);

	if (buflen == 0 || *fpos >= size) {
		read_unlock(&kclist_lock);
		return 0;
	}

	/* trim buflen to not go beyond EOF */
	if (buflen > size - *fpos)
		buflen = size - *fpos;

	/* construct an ELF core header if we'll need some of it */
	if (*fpos < elf_buflen) {
		char * elf_buf;

		tsz = elf_buflen - *fpos;
		if (buflen < tsz)
			tsz = buflen;
		elf_buf = kzalloc(elf_buflen, GFP_ATOMIC);
		if (!elf_buf) {
			read_unlock(&kclist_lock);
			return -ENOMEM;
		}
		elf_kcore_store_hdr(elf_buf, nphdr, elf_buflen);
		read_unlock(&kclist_lock);
		if (copy_to_user(buffer, elf_buf + *fpos, tsz)) {
			kfree(elf_buf);
			return -EFAULT;
		}
		kfree(elf_buf);
		buflen -= tsz;
		*fpos += tsz;
		buffer += tsz;
		acc += tsz;

		/* leave now if filled buffer already */
		if (buflen == 0)
			return acc;
	} else
		read_unlock(&kclist_lock);

	/*
	 * 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 - elf_buflen);
	if ((tsz = (PAGE_SIZE - (start & ~PAGE_MASK))) > buflen)
		tsz = buflen;
		
	while (buflen) {
		struct kcore_list *m;

		read_lock(&kclist_lock);
		list_for_each_entry(m, &kclist_head, list) {
			if (start >= m->addr && start < (m->addr+m->size))
				break;
		}
		read_unlock(&kclist_lock);

		if (&m->list == &kclist_head) {
			if (clear_user(buffer, tsz))
				return -EFAULT;
		} else if (is_vmalloc_or_module_addr((void *)start)) {
			char * elf_buf;

			elf_buf = kzalloc(tsz, GFP_KERNEL);
			if (!elf_buf)
				return -ENOMEM;
			vread(elf_buf, (char *)start, tsz);
			/* we have to zero-fill user buffer even if no read */
			if (copy_to_user(buffer, elf_buf, tsz)) {
				kfree(elf_buf);
				return -EFAULT;
			}
			kfree(elf_buf);
		} else {
			if (kern_addr_valid(start)) {
				unsigned long n;

				n = copy_to_user(buffer, (char *)start, tsz);
				/*
				 * We cannot distinguish between fault on source
				 * and fault on destination. When this happens
				 * we clear too and hope it will trigger the
				 * EFAULT again.
				 */
				if (n) { 
					if (clear_user(buffer + tsz - n,
								n))
						return -EFAULT;
				}
			} else {
				if (clear_user(buffer, tsz))
					return -EFAULT;
			}
		}
		buflen -= tsz;
		*fpos += tsz;
		buffer += tsz;
		acc += tsz;
		start += tsz;
		tsz = (buflen > PAGE_SIZE ? PAGE_SIZE : buflen);
	}

	return acc;
}


static int open_kcore(struct inode *inode, struct file *filp)
{
	if (!capable(CAP_SYS_RAWIO))
		return -EPERM;
	if (kcore_need_update)
		kcore_update_ram();
	if (i_size_read(inode) != proc_root_kcore->size) {
		mutex_lock(&inode->i_mutex);
		i_size_write(inode, proc_root_kcore->size);
		mutex_unlock(&inode->i_mutex);
	}
	return 0;
}


static const struct file_operations proc_kcore_operations = {
	.read		= read_kcore,
	.open		= open_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:
		write_lock(&kclist_lock);
		kcore_need_update = 1;
		write_unlock(&kclist_lock);
	}
	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
1da177e4 LT	1	/*
	2	* fs/proc/kcore.c kernel ELF core dumper
	3	*
	4	* Modelled on fs/exec.c:aout_core_dump()
	5	* Jeremy Fitzhardinge <[email protected]>
	6	* ELF version written by David Howells <[email protected]>
	7	* Modified and incorporated into 2.3.x by Tigran Aivazian <[email protected]>
	8	* Support to dump vmalloc'd areas (ELF only), Tigran Aivazian <[email protected]>
	9	* Safe accesses to vmalloc/direct-mapped discontiguous areas, Kanoj Sarcar <[email protected]>
	10	*/
	11
1da177e4 LT	12	#include <linux/mm.h>
1da177e4 LT	13	#include <linux/proc_fs.h>
2f96b8c1	14	#include <linux/kcore.h>
1da177e4	15	#include <linux/user.h>
16f7e0fe	16	#include <linux/capability.h>
1da177e4 LT	17	#include <linux/elf.h>
1da177e4 LT	18	#include <linux/elfcore.h>
3c743a7f	19	#include <linux/notifier.h>
1da177e4 LT	20	#include <linux/vmalloc.h>
1da177e4 LT	21	#include <linux/highmem.h>
87ebdc00	22	#include <linux/printk.h>
3089aa1b	23	#include <linux/bootmem.h>
1da177e4	24	#include <linux/init.h>
5a0e3ad6	25	#include <linux/slab.h>
1da177e4 LT	26	#include <asm/uaccess.h>
1da177e4 LT	27	#include <asm/io.h>
2ef43ec7	28	#include <linux/list.h>
3089aa1b	29	#include <linux/ioport.h>
3089aa1b	30	#include <linux/memory.h>
9492587c	31	#include <asm/sections.h>
59d8053f	32	#include "internal.h"
1da177e4	33
36027604	34	#define CORE_STR "CORE"
1da177e4	35
79885b22 EI	36	#ifndef ELF_CORE_EFLAGS
	37	#define ELF_CORE_EFLAGS 0
	38	#endif
	39
97ce5d6d AD	40	static struct proc_dir_entry *proc_root_kcore;
97ce5d6d AD	41
1da177e4 LT	42
	43	#ifndef kc_vaddr_to_offset
	44	#define kc_vaddr_to_offset(v) ((v) - PAGE_OFFSET)
	45	#endif
	46	#ifndef kc_offset_to_vaddr
	47	#define kc_offset_to_vaddr(o) ((o) + PAGE_OFFSET)
	48	#endif
	49
1da177e4 LT	50	/* An ELF note in memory */
	51	struct memelfnote
	52	{
	53	const char *name;
	54	int type;
	55	unsigned int datasz;
	56	void *data;
	57	};
	58
2ef43ec7	59	static LIST_HEAD(kclist_head);
1da177e4	60	static DEFINE_RWLOCK(kclist_lock);
3089aa1b	61	static int kcore_need_update = 1;
1da177e4 LT	62
1da177e4 LT	63	void
c30bb2a2	64	kclist_add(struct kcore_list new, void addr, size_t size, int type)
1da177e4 LT	65	{
	66	new->addr = (unsigned long)addr;
	67	new->size = size;
c30bb2a2	68	new->type = type;
1da177e4 LT	69
1da177e4 LT	70	write_lock(&kclist_lock);
2ef43ec7	71	list_add_tail(&new->list, &kclist_head);
1da177e4 LT	72	write_unlock(&kclist_lock);
	73	}
	74
	75	static size_t get_kcore_size(int nphdr, size_t elf_buflen)
	76	{
	77	size_t try, size;
	78	struct kcore_list *m;
	79
	80	nphdr = 1; / PT_NOTE */
	81	size = 0;
	82
2ef43ec7	83	list_for_each_entry(m, &kclist_head, list) {
1da177e4 LT	84	try = kc_vaddr_to_offset((size_t)m->addr + m->size);
	85	if (try > size)
	86	size = try;
	87	nphdr = nphdr + 1;
	88	}
	89	*elf_buflen = sizeof(struct elfhdr) +
	90	(nphdr + 2)sizeof(struct elf_phdr) +
36027604 MD	91	3 * ((sizeof(struct elf_note)) +
	92	roundup(sizeof(CORE_STR), 4)) +
	93	roundup(sizeof(struct elf_prstatus), 4) +
	94	roundup(sizeof(struct elf_prpsinfo), 4) +
5aaeb5c0	95	roundup(arch_task_struct_size, 4);
1da177e4 LT	96	elf_buflen = PAGE_ALIGN(elf_buflen);
	97	return size + *elf_buflen;
	98	}
	99
3089aa1b KH	100	static void free_kclist_ents(struct list_head *head)
	101	{
	102	struct kcore_list tmp, pos;
	103
	104	list_for_each_entry_safe(pos, tmp, head, list) {
	105	list_del(&pos->list);
	106	kfree(pos);
	107	}
	108	}
	109	/*
26562c59	110	* Replace all KCORE_RAM/KCORE_VMEMMAP information with passed list.
3089aa1b KH	111	*/
	112	static void __kcore_update_ram(struct list_head *list)
	113	{
678ad5d8 KH	114	int nphdr;
678ad5d8 KH	115	size_t size;
3089aa1b KH	116	struct kcore_list tmp, pos;
	117	LIST_HEAD(garbage);
	118
	119	write_lock(&kclist_lock);
	120	if (kcore_need_update) {
	121	list_for_each_entry_safe(pos, tmp, &kclist_head, list) {
26562c59 KH	122	if (pos->type == KCORE_RAM
26562c59 KH	123	\|\| pos->type == KCORE_VMEMMAP)
3089aa1b KH	124	list_move(&pos->list, &garbage);
	125	}
	126	list_splice_tail(list, &kclist_head);
	127	} else
	128	list_splice(list, &garbage);
	129	kcore_need_update = 0;
678ad5d8	130	proc_root_kcore->size = get_kcore_size(&nphdr, &size);
3089aa1b KH	131	write_unlock(&kclist_lock);
	132
	133	free_kclist_ents(&garbage);
	134	}
	135
	136
	137	#ifdef CONFIG_HIGHMEM
	138	/*
	139	* If no highmem, we can assume [0...max_low_pfn) continuous range of memory
	140	* because memory hole is not as big as !HIGHMEM case.
	141	* (HIGHMEM is special because part of memory is _invisible_ from the kernel.)
	142	*/
	143	static int kcore_update_ram(void)
	144	{
	145	LIST_HEAD(head);
	146	struct kcore_list *ent;
	147	int ret = 0;
	148
	149	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
	150	if (!ent)
	151	return -ENOMEM;
	152	ent->addr = (unsigned long)__va(0);
	153	ent->size = max_low_pfn << PAGE_SHIFT;
	154	ent->type = KCORE_RAM;
	155	list_add(&ent->list, &head);
	156	__kcore_update_ram(&head);
	157	return ret;
	158	}
	159
	160	#else /* !CONFIG_HIGHMEM */
	161
26562c59 KH	162	#ifdef CONFIG_SPARSEMEM_VMEMMAP
26562c59 KH	163	/* calculate vmemmap's address from given system ram pfn and register it */
b908243c DH	164	static int
b908243c DH	165	get_sparsemem_vmemmap_info(struct kcore_list ent, struct list_head head)
26562c59 KH	166	{
	167	unsigned long pfn = __pa(ent->addr) >> PAGE_SHIFT;
	168	unsigned long nr_pages = ent->size >> PAGE_SHIFT;
	169	unsigned long start, end;
	170	struct kcore_list vmm, tmp;
	171
	172
	173	start = ((unsigned long)pfn_to_page(pfn)) & PAGE_MASK;
	174	end = ((unsigned long)pfn_to_page(pfn + nr_pages)) - 1;
108a8a11	175	end = PAGE_ALIGN(end);
26562c59 KH	176	/* overlap check (because we have to align page */
	177	list_for_each_entry(tmp, head, list) {
	178	if (tmp->type != KCORE_VMEMMAP)
	179	continue;
	180	if (start < tmp->addr + tmp->size)
	181	if (end > tmp->addr)
	182	end = tmp->addr;
	183	}
	184	if (start < end) {
	185	vmm = kmalloc(sizeof(*vmm), GFP_KERNEL);
	186	if (!vmm)
	187	return 0;
	188	vmm->addr = start;
	189	vmm->size = end - start;
	190	vmm->type = KCORE_VMEMMAP;
	191	list_add_tail(&vmm->list, head);
	192	}
	193	return 1;
	194
	195	}
	196	#else
b908243c DH	197	static int
b908243c DH	198	get_sparsemem_vmemmap_info(struct kcore_list ent, struct list_head head)
26562c59 KH	199	{
	200	return 1;
	201	}
	202
	203	#endif
	204
3089aa1b KH	205	static int
	206	kclist_add_private(unsigned long pfn, unsigned long nr_pages, void *arg)
	207	{
	208	struct list_head head = (struct list_head )arg;
	209	struct kcore_list *ent;
	210
	211	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
	212	if (!ent)
	213	return -ENOMEM;
	214	ent->addr = (unsigned long)__va((pfn << PAGE_SHIFT));
	215	ent->size = nr_pages << PAGE_SHIFT;
	216
	217	/* Sanity check: Can happen in 32bit arch...maybe */
	218	if (ent->addr < (unsigned long) __va(0))
	219	goto free_out;
	220
	221	/* cut not-mapped area. ....from ppc-32 code. */
	222	if (ULONG_MAX - ent->addr < ent->size)
	223	ent->size = ULONG_MAX - ent->addr;
	224
	225	/* cut when vmalloc() area is higher than direct-map area */
	226	if (VMALLOC_START > (unsigned long)__va(0)) {
	227	if (ent->addr > VMALLOC_START)
	228	goto free_out;
	229	if (VMALLOC_START - ent->addr < ent->size)
	230	ent->size = VMALLOC_START - ent->addr;
	231	}
	232
	233	ent->type = KCORE_RAM;
	234	list_add_tail(&ent->list, head);
26562c59 KH	235
	236	if (!get_sparsemem_vmemmap_info(ent, head)) {
	237	list_del(&ent->list);
	238	goto free_out;
	239	}
	240
3089aa1b KH	241	return 0;
	242	free_out:
	243	kfree(ent);
	244	return 1;
	245	}
	246
	247	static int kcore_update_ram(void)
	248	{
	249	int nid, ret;
	250	unsigned long end_pfn;
	251	LIST_HEAD(head);
	252
	253	/* Not inialized....update now */
	254	/* find out "max pfn" */
	255	end_pfn = 0;
4ff1b2c2	256	for_each_node_state(nid, N_MEMORY) {
3089aa1b	257	unsigned long node_end;
83285c72	258	node_end = node_end_pfn(nid);
3089aa1b KH	259	if (end_pfn < node_end)
	260	end_pfn = node_end;
	261	}
	262	/* scan 0 to max_pfn */
	263	ret = walk_system_ram_range(0, end_pfn, &head, kclist_add_private);
	264	if (ret) {
	265	free_kclist_ents(&head);
	266	return -ENOMEM;
	267	}
	268	__kcore_update_ram(&head);
	269	return ret;
	270	}
	271	#endif /* CONFIG_HIGHMEM */
1da177e4 LT	272
	273	/*****************************************************************************/
	274	/*
	275	* determine size of ELF note
	276	*/
	277	static int notesize(struct memelfnote *en)
	278	{
	279	int sz;
	280
	281	sz = sizeof(struct elf_note);
632dd205	282	sz += roundup((strlen(en->name) + 1), 4);
1da177e4 LT	283	sz += roundup(en->datasz, 4);
	284
	285	return sz;
	286	} /* end notesize() */
	287
	288	/*****************************************************************************/
	289	/*
	290	* store a note in the header buffer
	291	*/
	292	static char storenote(struct memelfnote men, char *bufp)
	293	{
	294	struct elf_note en;
	295
	296	#define DUMP_WRITE(addr,nr) do { memcpy(bufp,addr,nr); bufp += nr; } while(0)
	297
632dd205	298	en.n_namesz = strlen(men->name) + 1;
1da177e4 LT	299	en.n_descsz = men->datasz;
	300	en.n_type = men->type;
	301
	302	DUMP_WRITE(&en, sizeof(en));
	303	DUMP_WRITE(men->name, en.n_namesz);
	304
	305	/* XXX - cast from long long to long to avoid need for libgcc.a */
	306	bufp = (char*) roundup((unsigned long)bufp,4);
	307	DUMP_WRITE(men->data, men->datasz);
	308	bufp = (char*) roundup((unsigned long)bufp,4);
	309
	310	#undef DUMP_WRITE
	311
	312	return bufp;
	313	} /* end storenote() */
	314
	315	/*
	316	* store an ELF coredump header in the supplied buffer
	317	* nphdr is the number of elf_phdr to insert
	318	*/
	319	static void elf_kcore_store_hdr(char *bufp, int nphdr, int dataoff)
	320	{
	321	struct elf_prstatus prstatus; /* NT_PRSTATUS */
	322	struct elf_prpsinfo prpsinfo; /* NT_PRPSINFO */
	323	struct elf_phdr nhdr, phdr;
	324	struct elfhdr *elf;
	325	struct memelfnote notes[3];
	326	off_t offset = 0;
	327	struct kcore_list *m;
	328
	329	/* setup ELF header */
	330	elf = (struct elfhdr *) bufp;
	331	bufp += sizeof(struct elfhdr);
	332	offset += sizeof(struct elfhdr);
	333	memcpy(elf->e_ident, ELFMAG, SELFMAG);
	334	elf->e_ident[EI_CLASS] = ELF_CLASS;
	335	elf->e_ident[EI_DATA] = ELF_DATA;
	336	elf->e_ident[EI_VERSION]= EV_CURRENT;
	337	elf->e_ident[EI_OSABI] = ELF_OSABI;
	338	memset(elf->e_ident+EI_PAD, 0, EI_NIDENT-EI_PAD);
	339	elf->e_type = ET_CORE;
	340	elf->e_machine = ELF_ARCH;
	341	elf->e_version = EV_CURRENT;
	342	elf->e_entry = 0;
	343	elf->e_phoff = sizeof(struct elfhdr);
	344	elf->e_shoff = 0;
79885b22	345	elf->e_flags = ELF_CORE_EFLAGS;
1da177e4 LT	346	elf->e_ehsize = sizeof(struct elfhdr);
	347	elf->e_phentsize= sizeof(struct elf_phdr);
	348	elf->e_phnum = nphdr;
	349	elf->e_shentsize= 0;
	350	elf->e_shnum = 0;
	351	elf->e_shstrndx = 0;
	352
	353	/* setup ELF PT_NOTE program header */
	354	nhdr = (struct elf_phdr *) bufp;
	355	bufp += sizeof(struct elf_phdr);
	356	offset += sizeof(struct elf_phdr);
	357	nhdr->p_type = PT_NOTE;
	358	nhdr->p_offset = 0;
	359	nhdr->p_vaddr = 0;
	360	nhdr->p_paddr = 0;
	361	nhdr->p_filesz = 0;
	362	nhdr->p_memsz = 0;
	363	nhdr->p_flags = 0;
	364	nhdr->p_align = 0;
	365
	366	/* setup ELF PT_LOAD program header for every area */
2ef43ec7	367	list_for_each_entry(m, &kclist_head, list) {
1da177e4 LT	368	phdr = (struct elf_phdr *) bufp;
	369	bufp += sizeof(struct elf_phdr);
	370	offset += sizeof(struct elf_phdr);
	371
	372	phdr->p_type = PT_LOAD;
	373	phdr->p_flags = PF_R\|PF_W\|PF_X;
	374	phdr->p_offset = kc_vaddr_to_offset(m->addr) + dataoff;
	375	phdr->p_vaddr = (size_t)m->addr;
	376	phdr->p_paddr = 0;
	377	phdr->p_filesz = phdr->p_memsz = m->size;
	378	phdr->p_align = PAGE_SIZE;
	379	}
	380
	381	/*
	382	* Set up the notes in similar form to SVR4 core dumps made
	383	* with info from their /proc.
	384	*/
	385	nhdr->p_offset = offset;
	386
	387	/* set up the process status */
36027604	388	notes[0].name = CORE_STR;
1da177e4 LT	389	notes[0].type = NT_PRSTATUS;
	390	notes[0].datasz = sizeof(struct elf_prstatus);
	391	notes[0].data = &prstatus;
	392
	393	memset(&prstatus, 0, sizeof(struct elf_prstatus));
	394
	395	nhdr->p_filesz = notesize(&notes[0]);
	396	bufp = storenote(&notes[0], bufp);
	397
	398	/* set up the process info */
36027604	399	notes[1].name = CORE_STR;
1da177e4 LT	400	notes[1].type = NT_PRPSINFO;
	401	notes[1].datasz = sizeof(struct elf_prpsinfo);
	402	notes[1].data = &prpsinfo;
	403
	404	memset(&prpsinfo, 0, sizeof(struct elf_prpsinfo));
	405	prpsinfo.pr_state = 0;
	406	prpsinfo.pr_sname = 'R';
	407	prpsinfo.pr_zomb = 0;
	408
	409	strcpy(prpsinfo.pr_fname, "vmlinux");
30bc30df	410	strlcpy(prpsinfo.pr_psargs, saved_command_line, sizeof(prpsinfo.pr_psargs));
1da177e4 LT	411
	412	nhdr->p_filesz += notesize(&notes[1]);
	413	bufp = storenote(&notes[1], bufp);
	414
	415	/* set up the task structure */
36027604	416	notes[2].name = CORE_STR;
1da177e4	417	notes[2].type = NT_TASKSTRUCT;
5aaeb5c0	418	notes[2].datasz = arch_task_struct_size;
1da177e4 LT	419	notes[2].data = current;
	420
	421	nhdr->p_filesz += notesize(&notes[2]);
	422	bufp = storenote(&notes[2], bufp);
	423
	424	} /* end elf_kcore_store_hdr() */
	425
	426	/*****************************************************************************/
	427	/*
	428	* read from the ELF header and then kernel memory
	429	*/
	430	static ssize_t
	431	read_kcore(struct file file, char __user buffer, size_t buflen, loff_t *fpos)
	432	{
	433	ssize_t acc = 0;
	434	size_t size, tsz;
	435	size_t elf_buflen;
	436	int nphdr;
	437	unsigned long start;
	438
	439	read_lock(&kclist_lock);
678ad5d8 KH	440	size = get_kcore_size(&nphdr, &elf_buflen);
678ad5d8 KH	441
1da177e4 LT	442	if (buflen == 0 \|\| *fpos >= size) {
	443	read_unlock(&kclist_lock);
	444	return 0;
	445	}
	446
	447	/* trim buflen to not go beyond EOF */
	448	if (buflen > size - *fpos)
	449	buflen = size - *fpos;
	450
	451	/* construct an ELF core header if we'll need some of it */
	452	if (*fpos < elf_buflen) {
	453	char * elf_buf;
	454
	455	tsz = elf_buflen - *fpos;
	456	if (buflen < tsz)
	457	tsz = buflen;
f8314dc6	458	elf_buf = kzalloc(elf_buflen, GFP_ATOMIC);
1da177e4 LT	459	if (!elf_buf) {
	460	read_unlock(&kclist_lock);
	461	return -ENOMEM;
	462	}
1da177e4 LT	463	elf_kcore_store_hdr(elf_buf, nphdr, elf_buflen);
	464	read_unlock(&kclist_lock);
	465	if (copy_to_user(buffer, elf_buf + *fpos, tsz)) {
	466	kfree(elf_buf);
	467	return -EFAULT;
	468	}
	469	kfree(elf_buf);
	470	buflen -= tsz;
	471	*fpos += tsz;
	472	buffer += tsz;
	473	acc += tsz;
	474
	475	/* leave now if filled buffer already */
	476	if (buflen == 0)
	477	return acc;
	478	} else
	479	read_unlock(&kclist_lock);
	480
	481	/*
	482	* Check to see if our file offset matches with any of
	483	* the addresses in the elf_phdr on our list.
	484	*/
	485	start = kc_offset_to_vaddr(*fpos - elf_buflen);
	486	if ((tsz = (PAGE_SIZE - (start & ~PAGE_MASK))) > buflen)
	487	tsz = buflen;
	488
	489	while (buflen) {
	490	struct kcore_list *m;
	491
	492	read_lock(&kclist_lock);
2ef43ec7	493	list_for_each_entry(m, &kclist_head, list) {
1da177e4 LT	494	if (start >= m->addr && start < (m->addr+m->size))
	495	break;
	496	}
	497	read_unlock(&kclist_lock);
	498
4fd2c20d	499	if (&m->list == &kclist_head) {
1da177e4 LT	500	if (clear_user(buffer, tsz))
1da177e4 LT	501	return -EFAULT;
81ac3ad9	502	} else if (is_vmalloc_or_module_addr((void *)start)) {
1da177e4	503	char * elf_buf;
1da177e4	504
f8314dc6	505	elf_buf = kzalloc(tsz, GFP_KERNEL);
1da177e4 LT	506	if (!elf_buf)
1da177e4 LT	507	return -ENOMEM;
73d7c33e KH	508	vread(elf_buf, (char *)start, tsz);
73d7c33e KH	509	/* we have to zero-fill user buffer even if no read */
1da177e4 LT	510	if (copy_to_user(buffer, elf_buf, tsz)) {
	511	kfree(elf_buf);
	512	return -EFAULT;
	513	}
	514	kfree(elf_buf);
	515	} else {
	516	if (kern_addr_valid(start)) {
	517	unsigned long n;
	518
	519	n = copy_to_user(buffer, (char *)start, tsz);
	520	/*
f1f996b6	521	* We cannot distinguish between fault on source
1da177e4 LT	522	* and fault on destination. When this happens
	523	* we clear too and hope it will trigger the
	524	* EFAULT again.
	525	*/
	526	if (n) {
	527	if (clear_user(buffer + tsz - n,
0635170b	528	n))
1da177e4 LT	529	return -EFAULT;
	530	}
	531	} else {
	532	if (clear_user(buffer, tsz))
	533	return -EFAULT;
	534	}
	535	}
	536	buflen -= tsz;
	537	*fpos += tsz;
	538	buffer += tsz;
	539	acc += tsz;
	540	start += tsz;
	541	tsz = (buflen > PAGE_SIZE ? PAGE_SIZE : buflen);
	542	}
	543
	544	return acc;
	545	}
97ce5d6d	546
3089aa1b KH	547
	548	static int open_kcore(struct inode inode, struct file filp)
	549	{
	550	if (!capable(CAP_SYS_RAWIO))
	551	return -EPERM;
	552	if (kcore_need_update)
	553	kcore_update_ram();
0d4c36a9 KH	554	if (i_size_read(inode) != proc_root_kcore->size) {
	555	mutex_lock(&inode->i_mutex);
	556	i_size_write(inode, proc_root_kcore->size);
	557	mutex_unlock(&inode->i_mutex);
	558	}
3089aa1b KH	559	return 0;
	560	}
	561
	562
	563	static const struct file_operations proc_kcore_operations = {
	564	.read = read_kcore,
	565	.open = open_kcore,
ceff1a77	566	.llseek = default_llseek,
3089aa1b KH	567	};
3089aa1b KH	568
3089aa1b KH	569	/* just remember that we have to update kcore */
	570	static int __meminit kcore_callback(struct notifier_block *self,
	571	unsigned long action, void *arg)
	572	{
	573	switch (action) {
	574	case MEM_ONLINE:
	575	case MEM_OFFLINE:
	576	write_lock(&kclist_lock);
	577	kcore_need_update = 1;
	578	write_unlock(&kclist_lock);
	579	}
	580	return NOTIFY_OK;
	581	}
3089aa1b	582
3c743a7f AM	583	static struct notifier_block kcore_callback_nb __meminitdata = {
	584	.notifier_call = kcore_callback,
	585	.priority = 0,
	586	};
3089aa1b	587
a0614da8 KH	588	static struct kcore_list kcore_vmalloc;
a0614da8 KH	589
9492587c KH	590	#ifdef CONFIG_ARCH_PROC_KCORE_TEXT
	591	static struct kcore_list kcore_text;
	592	/*
	593	* If defined, special segment is used for mapping kernel text instead of
	594	* direct-map area. We need to create special TEXT section.
	595	*/
	596	static void __init proc_kcore_text_init(void)
	597	{
36e15263	598	kclist_add(&kcore_text, _text, _end - _text, KCORE_TEXT);
9492587c KH	599	}
	600	#else
	601	static void __init proc_kcore_text_init(void)
	602	{
	603	}
	604	#endif
	605
81ac3ad9 KH	606	#if defined(CONFIG_MODULES) && defined(MODULES_VADDR)
	607	/*
	608	* MODULES_VADDR has no intersection with VMALLOC_ADDR.
	609	*/
	610	struct kcore_list kcore_modules;
	611	static void __init add_modules_range(void)
	612	{
bf3e2692 BH	613	if (MODULES_VADDR != VMALLOC_START && MODULES_END != VMALLOC_END) {
bf3e2692 BH	614	kclist_add(&kcore_modules, (void *)MODULES_VADDR,
81ac3ad9	615	MODULES_END - MODULES_VADDR, KCORE_VMALLOC);
bf3e2692	616	}
81ac3ad9 KH	617	}
	618	#else
	619	static void __init add_modules_range(void)
	620	{
	621	}
	622	#endif
	623
97ce5d6d AD	624	static int __init proc_kcore_init(void)
97ce5d6d AD	625	{
3089aa1b KH	626	proc_root_kcore = proc_create("kcore", S_IRUSR, NULL,
3089aa1b KH	627	&proc_kcore_operations);
90396f96	628	if (!proc_root_kcore) {
87ebdc00	629	pr_err("couldn't create /proc/kcore\n");
90396f96 KH	630	return 0; /* Always returns 0. */
90396f96 KH	631	}
3089aa1b	632	/* Store text area if it's special */
9492587c	633	proc_kcore_text_init();
3089aa1b	634	/* Store vmalloc area */
a0614da8 KH	635	kclist_add(&kcore_vmalloc, (void *)VMALLOC_START,
a0614da8 KH	636	VMALLOC_END - VMALLOC_START, KCORE_VMALLOC);
81ac3ad9	637	add_modules_range();
3089aa1b KH	638	/* Store direct-map area from physical memory map */
3089aa1b KH	639	kcore_update_ram();
3c743a7f	640	register_hotmemory_notifier(&kcore_callback_nb);
3089aa1b	641
97ce5d6d AD	642	return 0;
97ce5d6d AD	643	}
abaf3787	644	fs_initcall(proc_kcore_init);