[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/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 */
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
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 distingush 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 <[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>
	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
	159	/* calculate vmemmap's address from given system ram pfn and register it */
	160	int get_sparsemem_vmemmap_info(struct kcore_list ent, struct list_head head)
	161	{
	162	unsigned long pfn = __pa(ent->addr) >> PAGE_SHIFT;
	163	unsigned long nr_pages = ent->size >> PAGE_SHIFT;
	164	unsigned long start, end;
	165	struct kcore_list vmm, tmp;
	166
	167
	168	start = ((unsigned long)pfn_to_page(pfn)) & PAGE_MASK;
	169	end = ((unsigned long)pfn_to_page(pfn + nr_pages)) - 1;
	170	end = ALIGN(end, PAGE_SIZE);
	171	/* overlap check (because we have to align page */
	172	list_for_each_entry(tmp, head, list) {
	173	if (tmp->type != KCORE_VMEMMAP)
	174	continue;
	175	if (start < tmp->addr + tmp->size)
	176	if (end > tmp->addr)
	177	end = tmp->addr;
	178	}
	179	if (start < end) {
	180	vmm = kmalloc(sizeof(*vmm), GFP_KERNEL);
	181	if (!vmm)
	182	return 0;
	183	vmm->addr = start;
	184	vmm->size = end - start;
	185	vmm->type = KCORE_VMEMMAP;
	186	list_add_tail(&vmm->list, head);
	187	}
	188	return 1;
	189
	190	}
	191	#else
	192	int get_sparsemem_vmemmap_info(struct kcore_list ent, struct list_head head)
	193	{
	194	return 1;
	195	}
	196
	197	#endif
	198
3089aa1b KH	199	static int
	200	kclist_add_private(unsigned long pfn, unsigned long nr_pages, void *arg)
	201	{
	202	struct list_head head = (struct list_head )arg;
	203	struct kcore_list *ent;
	204
	205	ent = kmalloc(sizeof(*ent), GFP_KERNEL);
	206	if (!ent)
	207	return -ENOMEM;
	208	ent->addr = (unsigned long)__va((pfn << PAGE_SHIFT));
	209	ent->size = nr_pages << PAGE_SHIFT;
	210
	211	/* Sanity check: Can happen in 32bit arch...maybe */
	212	if (ent->addr < (unsigned long) __va(0))
	213	goto free_out;
	214
	215	/* cut not-mapped area. ....from ppc-32 code. */
	216	if (ULONG_MAX - ent->addr < ent->size)
	217	ent->size = ULONG_MAX - ent->addr;
	218
	219	/* cut when vmalloc() area is higher than direct-map area */
	220	if (VMALLOC_START > (unsigned long)__va(0)) {
	221	if (ent->addr > VMALLOC_START)
	222	goto free_out;
	223	if (VMALLOC_START - ent->addr < ent->size)
	224	ent->size = VMALLOC_START - ent->addr;
	225	}
	226
	227	ent->type = KCORE_RAM;
	228	list_add_tail(&ent->list, head);
26562c59 KH	229
	230	if (!get_sparsemem_vmemmap_info(ent, head)) {
	231	list_del(&ent->list);
	232	goto free_out;
	233	}
	234
3089aa1b KH	235	return 0;
	236	free_out:
	237	kfree(ent);
	238	return 1;
	239	}
	240
	241	static int kcore_update_ram(void)
	242	{
	243	int nid, ret;
	244	unsigned long end_pfn;
	245	LIST_HEAD(head);
	246
	247	/* Not inialized....update now */
	248	/* find out "max pfn" */
	249	end_pfn = 0;
	250	for_each_node_state(nid, N_HIGH_MEMORY) {
	251	unsigned long node_end;
	252	node_end = NODE_DATA(nid)->node_start_pfn +
	253	NODE_DATA(nid)->node_spanned_pages;
	254	if (end_pfn < node_end)
	255	end_pfn = node_end;
	256	}
	257	/* scan 0 to max_pfn */
	258	ret = walk_system_ram_range(0, end_pfn, &head, kclist_add_private);
	259	if (ret) {
	260	free_kclist_ents(&head);
	261	return -ENOMEM;
	262	}
	263	__kcore_update_ram(&head);
	264	return ret;
	265	}
	266	#endif /* CONFIG_HIGHMEM */
1da177e4 LT	267
	268	/*****************************************************************************/
	269	/*
	270	* determine size of ELF note
	271	*/
	272	static int notesize(struct memelfnote *en)
	273	{
	274	int sz;
	275
	276	sz = sizeof(struct elf_note);
632dd205	277	sz += roundup((strlen(en->name) + 1), 4);
1da177e4 LT	278	sz += roundup(en->datasz, 4);
	279
	280	return sz;
	281	} /* end notesize() */
	282
	283	/*****************************************************************************/
	284	/*
	285	* store a note in the header buffer
	286	*/
	287	static char storenote(struct memelfnote men, char *bufp)
	288	{
	289	struct elf_note en;
	290
	291	#define DUMP_WRITE(addr,nr) do { memcpy(bufp,addr,nr); bufp += nr; } while(0)
	292
632dd205	293	en.n_namesz = strlen(men->name) + 1;
1da177e4 LT	294	en.n_descsz = men->datasz;
	295	en.n_type = men->type;
	296
	297	DUMP_WRITE(&en, sizeof(en));
	298	DUMP_WRITE(men->name, en.n_namesz);
	299
	300	/* XXX - cast from long long to long to avoid need for libgcc.a */
	301	bufp = (char*) roundup((unsigned long)bufp,4);
	302	DUMP_WRITE(men->data, men->datasz);
	303	bufp = (char*) roundup((unsigned long)bufp,4);
	304
	305	#undef DUMP_WRITE
	306
	307	return bufp;
	308	} /* end storenote() */
	309
	310	/*
	311	* store an ELF coredump header in the supplied buffer
	312	* nphdr is the number of elf_phdr to insert
	313	*/
	314	static void elf_kcore_store_hdr(char *bufp, int nphdr, int dataoff)
	315	{
	316	struct elf_prstatus prstatus; /* NT_PRSTATUS */
	317	struct elf_prpsinfo prpsinfo; /* NT_PRPSINFO */
	318	struct elf_phdr nhdr, phdr;
	319	struct elfhdr *elf;
	320	struct memelfnote notes[3];
	321	off_t offset = 0;
	322	struct kcore_list *m;
	323
	324	/* setup ELF header */
	325	elf = (struct elfhdr *) bufp;
	326	bufp += sizeof(struct elfhdr);
	327	offset += sizeof(struct elfhdr);
	328	memcpy(elf->e_ident, ELFMAG, SELFMAG);
	329	elf->e_ident[EI_CLASS] = ELF_CLASS;
	330	elf->e_ident[EI_DATA] = ELF_DATA;
	331	elf->e_ident[EI_VERSION]= EV_CURRENT;
	332	elf->e_ident[EI_OSABI] = ELF_OSABI;
	333	memset(elf->e_ident+EI_PAD, 0, EI_NIDENT-EI_PAD);
	334	elf->e_type = ET_CORE;
	335	elf->e_machine = ELF_ARCH;
	336	elf->e_version = EV_CURRENT;
	337	elf->e_entry = 0;
	338	elf->e_phoff = sizeof(struct elfhdr);
	339	elf->e_shoff = 0;
79885b22	340	elf->e_flags = ELF_CORE_EFLAGS;
1da177e4 LT	341	elf->e_ehsize = sizeof(struct elfhdr);
	342	elf->e_phentsize= sizeof(struct elf_phdr);
	343	elf->e_phnum = nphdr;
	344	elf->e_shentsize= 0;
	345	elf->e_shnum = 0;
	346	elf->e_shstrndx = 0;
	347
	348	/* setup ELF PT_NOTE program header */
	349	nhdr = (struct elf_phdr *) bufp;
	350	bufp += sizeof(struct elf_phdr);
	351	offset += sizeof(struct elf_phdr);
	352	nhdr->p_type = PT_NOTE;
	353	nhdr->p_offset = 0;
	354	nhdr->p_vaddr = 0;
	355	nhdr->p_paddr = 0;
	356	nhdr->p_filesz = 0;
	357	nhdr->p_memsz = 0;
	358	nhdr->p_flags = 0;
	359	nhdr->p_align = 0;
	360
	361	/* setup ELF PT_LOAD program header for every area */
2ef43ec7	362	list_for_each_entry(m, &kclist_head, list) {
1da177e4 LT	363	phdr = (struct elf_phdr *) bufp;
	364	bufp += sizeof(struct elf_phdr);
	365	offset += sizeof(struct elf_phdr);
	366
	367	phdr->p_type = PT_LOAD;
	368	phdr->p_flags = PF_R\|PF_W\|PF_X;
	369	phdr->p_offset = kc_vaddr_to_offset(m->addr) + dataoff;
	370	phdr->p_vaddr = (size_t)m->addr;
	371	phdr->p_paddr = 0;
	372	phdr->p_filesz = phdr->p_memsz = m->size;
	373	phdr->p_align = PAGE_SIZE;
	374	}
	375
	376	/*
	377	* Set up the notes in similar form to SVR4 core dumps made
	378	* with info from their /proc.
	379	*/
	380	nhdr->p_offset = offset;
	381
	382	/* set up the process status */
36027604	383	notes[0].name = CORE_STR;
1da177e4 LT	384	notes[0].type = NT_PRSTATUS;
	385	notes[0].datasz = sizeof(struct elf_prstatus);
	386	notes[0].data = &prstatus;
	387
	388	memset(&prstatus, 0, sizeof(struct elf_prstatus));
	389
	390	nhdr->p_filesz = notesize(&notes[0]);
	391	bufp = storenote(&notes[0], bufp);
	392
	393	/* set up the process info */
36027604	394	notes[1].name = CORE_STR;
1da177e4 LT	395	notes[1].type = NT_PRPSINFO;
	396	notes[1].datasz = sizeof(struct elf_prpsinfo);
	397	notes[1].data = &prpsinfo;
	398
	399	memset(&prpsinfo, 0, sizeof(struct elf_prpsinfo));
	400	prpsinfo.pr_state = 0;
	401	prpsinfo.pr_sname = 'R';
	402	prpsinfo.pr_zomb = 0;
	403
	404	strcpy(prpsinfo.pr_fname, "vmlinux");
	405	strncpy(prpsinfo.pr_psargs, saved_command_line, ELF_PRARGSZ);
	406
	407	nhdr->p_filesz += notesize(&notes[1]);
	408	bufp = storenote(&notes[1], bufp);
	409
	410	/* set up the task structure */
36027604	411	notes[2].name = CORE_STR;
1da177e4 LT	412	notes[2].type = NT_TASKSTRUCT;
	413	notes[2].datasz = sizeof(struct task_struct);
	414	notes[2].data = current;
	415
	416	nhdr->p_filesz += notesize(&notes[2]);
	417	bufp = storenote(&notes[2], bufp);
	418
	419	} /* end elf_kcore_store_hdr() */
	420
	421	/*****************************************************************************/
	422	/*
	423	* read from the ELF header and then kernel memory
	424	*/
	425	static ssize_t
	426	read_kcore(struct file file, char __user buffer, size_t buflen, loff_t *fpos)
	427	{
	428	ssize_t acc = 0;
	429	size_t size, tsz;
	430	size_t elf_buflen;
	431	int nphdr;
	432	unsigned long start;
	433
	434	read_lock(&kclist_lock);
678ad5d8 KH	435	size = get_kcore_size(&nphdr, &elf_buflen);
678ad5d8 KH	436
1da177e4 LT	437	if (buflen == 0 \|\| *fpos >= size) {
	438	read_unlock(&kclist_lock);
	439	return 0;
	440	}
	441
	442	/* trim buflen to not go beyond EOF */
	443	if (buflen > size - *fpos)
	444	buflen = size - *fpos;
	445
	446	/* construct an ELF core header if we'll need some of it */
	447	if (*fpos < elf_buflen) {
	448	char * elf_buf;
	449
	450	tsz = elf_buflen - *fpos;
	451	if (buflen < tsz)
	452	tsz = buflen;
f8314dc6	453	elf_buf = kzalloc(elf_buflen, GFP_ATOMIC);
1da177e4 LT	454	if (!elf_buf) {
	455	read_unlock(&kclist_lock);
	456	return -ENOMEM;
	457	}
1da177e4 LT	458	elf_kcore_store_hdr(elf_buf, nphdr, elf_buflen);
	459	read_unlock(&kclist_lock);
	460	if (copy_to_user(buffer, elf_buf + *fpos, tsz)) {
	461	kfree(elf_buf);
	462	return -EFAULT;
	463	}
	464	kfree(elf_buf);
	465	buflen -= tsz;
	466	*fpos += tsz;
	467	buffer += tsz;
	468	acc += tsz;
	469
	470	/* leave now if filled buffer already */
	471	if (buflen == 0)
	472	return acc;
	473	} else
	474	read_unlock(&kclist_lock);
	475
	476	/*
	477	* Check to see if our file offset matches with any of
	478	* the addresses in the elf_phdr on our list.
	479	*/
	480	start = kc_offset_to_vaddr(*fpos - elf_buflen);
	481	if ((tsz = (PAGE_SIZE - (start & ~PAGE_MASK))) > buflen)
	482	tsz = buflen;
	483
	484	while (buflen) {
	485	struct kcore_list *m;
	486
	487	read_lock(&kclist_lock);
2ef43ec7	488	list_for_each_entry(m, &kclist_head, list) {
1da177e4 LT	489	if (start >= m->addr && start < (m->addr+m->size))
	490	break;
	491	}
	492	read_unlock(&kclist_lock);
	493
4fd2c20d	494	if (&m->list == &kclist_head) {
1da177e4 LT	495	if (clear_user(buffer, tsz))
1da177e4 LT	496	return -EFAULT;
81ac3ad9	497	} else if (is_vmalloc_or_module_addr((void *)start)) {
1da177e4	498	char * elf_buf;
1da177e4	499
f8314dc6	500	elf_buf = kzalloc(tsz, GFP_KERNEL);
1da177e4 LT	501	if (!elf_buf)
1da177e4 LT	502	return -ENOMEM;
73d7c33e KH	503	vread(elf_buf, (char *)start, tsz);
73d7c33e KH	504	/* we have to zero-fill user buffer even if no read */
1da177e4 LT	505	if (copy_to_user(buffer, elf_buf, tsz)) {
	506	kfree(elf_buf);
	507	return -EFAULT;
	508	}
	509	kfree(elf_buf);
	510	} else {
	511	if (kern_addr_valid(start)) {
	512	unsigned long n;
	513
	514	n = copy_to_user(buffer, (char *)start, tsz);
	515	/*
	516	* We cannot distingush between fault on source
	517	* and fault on destination. When this happens
	518	* we clear too and hope it will trigger the
	519	* EFAULT again.
	520	*/
	521	if (n) {
	522	if (clear_user(buffer + tsz - n,
0635170b	523	n))
1da177e4 LT	524	return -EFAULT;
	525	}
	526	} else {
	527	if (clear_user(buffer, tsz))
	528	return -EFAULT;
	529	}
	530	}
	531	buflen -= tsz;
	532	*fpos += tsz;
	533	buffer += tsz;
	534	acc += tsz;
	535	start += tsz;
	536	tsz = (buflen > PAGE_SIZE ? PAGE_SIZE : buflen);
	537	}
	538
	539	return acc;
	540	}
97ce5d6d	541
3089aa1b KH	542
	543	static int open_kcore(struct inode inode, struct file filp)
	544	{
	545	if (!capable(CAP_SYS_RAWIO))
	546	return -EPERM;
	547	if (kcore_need_update)
	548	kcore_update_ram();
0d4c36a9 KH	549	if (i_size_read(inode) != proc_root_kcore->size) {
	550	mutex_lock(&inode->i_mutex);
	551	i_size_write(inode, proc_root_kcore->size);
	552	mutex_unlock(&inode->i_mutex);
	553	}
3089aa1b KH	554	return 0;
	555	}
	556
	557
	558	static const struct file_operations proc_kcore_operations = {
	559	.read = read_kcore,
	560	.open = open_kcore,
ceff1a77	561	.llseek = default_llseek,
3089aa1b KH	562	};
	563
	564	#ifdef CONFIG_MEMORY_HOTPLUG
	565	/* just remember that we have to update kcore */
	566	static int __meminit kcore_callback(struct notifier_block *self,
	567	unsigned long action, void *arg)
	568	{
	569	switch (action) {
	570	case MEM_ONLINE:
	571	case MEM_OFFLINE:
	572	write_lock(&kclist_lock);
	573	kcore_need_update = 1;
	574	write_unlock(&kclist_lock);
	575	}
	576	return NOTIFY_OK;
	577	}
	578	#endif
	579
	580
a0614da8 KH	581	static struct kcore_list kcore_vmalloc;
a0614da8 KH	582
9492587c KH	583	#ifdef CONFIG_ARCH_PROC_KCORE_TEXT
	584	static struct kcore_list kcore_text;
	585	/*
	586	* If defined, special segment is used for mapping kernel text instead of
	587	* direct-map area. We need to create special TEXT section.
	588	*/
	589	static void __init proc_kcore_text_init(void)
	590	{
36e15263	591	kclist_add(&kcore_text, _text, _end - _text, KCORE_TEXT);
9492587c KH	592	}
	593	#else
	594	static void __init proc_kcore_text_init(void)
	595	{
	596	}
	597	#endif
	598
81ac3ad9 KH	599	#if defined(CONFIG_MODULES) && defined(MODULES_VADDR)
	600	/*
	601	* MODULES_VADDR has no intersection with VMALLOC_ADDR.
	602	*/
	603	struct kcore_list kcore_modules;
	604	static void __init add_modules_range(void)
	605	{
	606	kclist_add(&kcore_modules, (void *)MODULES_VADDR,
	607	MODULES_END - MODULES_VADDR, KCORE_VMALLOC);
	608	}
	609	#else
	610	static void __init add_modules_range(void)
	611	{
	612	}
	613	#endif
	614
97ce5d6d AD	615	static int __init proc_kcore_init(void)
97ce5d6d AD	616	{
3089aa1b KH	617	proc_root_kcore = proc_create("kcore", S_IRUSR, NULL,
3089aa1b KH	618	&proc_kcore_operations);
90396f96 KH	619	if (!proc_root_kcore) {
	620	printk(KERN_ERR "couldn't create /proc/kcore\n");
	621	return 0; /* Always returns 0. */
	622	}
3089aa1b	623	/* Store text area if it's special */
9492587c	624	proc_kcore_text_init();
3089aa1b	625	/* Store vmalloc area */
a0614da8 KH	626	kclist_add(&kcore_vmalloc, (void *)VMALLOC_START,
a0614da8 KH	627	VMALLOC_END - VMALLOC_START, KCORE_VMALLOC);
81ac3ad9	628	add_modules_range();
3089aa1b KH	629	/* Store direct-map area from physical memory map */
	630	kcore_update_ram();
	631	hotplug_memory_notifier(kcore_callback, 0);
3089aa1b	632
97ce5d6d AD	633	return 0;
	634	}
	635	module_init(proc_kcore_init);