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

/*
 *	fs/proc/kcore.c kernel ELF core dumper
 *
 *	Modelled on fs/exec.c:aout_core_dump()
 *	Jeremy Fitzhardinge <jeremy@sw.oz.au>
 *	ELF version written by David Howells <David.Howells@nexor.co.uk>
 *	Modified and incorporated into 2.3.x by Tigran Aivazian <tigran@veritas.com>
 *	Support to dump vmalloc'd areas (ELF only), Tigran Aivazian <tigran@veritas.com>
 *	Safe accesses to vmalloc/direct-mapped discontiguous areas, Kanoj Sarcar <kanoj@sgi.com>
 */

#include <linux/mm.h>
#include <linux/proc_fs.h>
#include <linux/user.h>
#include <linux/capability.h>
#include <linux/elf.h>
#include <linux/elfcore.h>
#include <linux/vmalloc.h>
#include <linux/highmem.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>

#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(sizeof(struct task_struct), 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 = ALIGN(end, PAGE_SIZE);
	/* 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_HIGH_MEMORY) {
		unsigned long node_end;
		node_end  = NODE_DATA(nid)->node_start_pfn +
			NODE_DATA(nid)->node_spanned_pages;
		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");
	strncpy(prpsinfo.pr_psargs, saved_command_line, ELF_PRARGSZ);

	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	= sizeof(struct task_struct);
	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,
};

#ifdef CONFIG_MEMORY_HOTPLUG
/* 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;
}
#endif


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)
{
	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) {
		printk(KERN_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();
	hotplug_memory_notifier(kcore_callback, 0);

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