[net-next-2.6.git] / drivers / char / mspec.c

/*
 * Copyright (C) 2001-2006 Silicon Graphics, Inc.  All rights
 * reserved.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of version 2 of the GNU General Public License
 * as published by the Free Software Foundation.
 */

/*
 * SN Platform Special Memory (mspec) Support
 *
 * This driver exports the SN special memory (mspec) facility to user
 * processes.
 * There are three types of memory made available thru this driver:
 * fetchops, uncached and cached.
 *
 * Fetchops are atomic memory operations that are implemented in the
 * memory controller on SGI SN hardware.
 *
 * Uncached are used for memory write combining feature of the ia64
 * cpu.
 *
 * Cached are used for areas of memory that are used as cached addresses
 * on our partition and used as uncached addresses from other partitions.
 * Due to a design constraint of the SN2 Shub, you can not have processors
 * on the same FSB perform both a cached and uncached reference to the
 * same cache line.  These special memory cached regions prevent the
 * kernel from ever dropping in a TLB entry and therefore prevent the
 * processor from ever speculating a cache line from this page.
 */

#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/errno.h>
#include <linux/miscdevice.h>
#include <linux/spinlock.h>
#include <linux/mm.h>
#include <linux/fs.h>
#include <linux/vmalloc.h>
#include <linux/string.h>
#include <linux/slab.h>
#include <linux/numa.h>
#include <asm/page.h>
#include <asm/system.h>
#include <asm/pgtable.h>
#include <asm/atomic.h>
#include <asm/tlbflush.h>
#include <asm/uncached.h>
#include <asm/sn/addrs.h>
#include <asm/sn/arch.h>
#include <asm/sn/mspec.h>
#include <asm/sn/sn_cpuid.h>
#include <asm/sn/io.h>
#include <asm/sn/bte.h>
#include <asm/sn/shubio.h>


#define FETCHOP_ID	"SGI Fetchop,"
#define CACHED_ID	"Cached,"
#define UNCACHED_ID	"Uncached"
#define REVISION	"4.0"
#define MSPEC_BASENAME	"mspec"

/*
 * Page types allocated by the device.
 */
enum mspec_page_type {
	MSPEC_FETCHOP = 1,
	MSPEC_CACHED,
	MSPEC_UNCACHED
};

#ifdef CONFIG_SGI_SN
static int is_sn2;
#else
#define is_sn2		0
#endif

/*
 * One of these structures is allocated when an mspec region is mmaped. The
 * structure is pointed to by the vma->vm_private_data field in the vma struct.
 * This structure is used to record the addresses of the mspec pages.
 * This structure is shared by all vma's that are split off from the
 * original vma when split_vma()'s are done.
 *
 * The refcnt is incremented atomically because mm->mmap_sem does not
 * protect in fork case where multiple tasks share the vma_data.
 */
struct vma_data {
	atomic_t refcnt;	/* Number of vmas sharing the data. */
	spinlock_t lock;	/* Serialize access to this structure. */
	int count;		/* Number of pages allocated. */
	enum mspec_page_type type; /* Type of pages allocated. */
	int flags;		/* See VMD_xxx below. */
	unsigned long vm_start;	/* Original (unsplit) base. */
	unsigned long vm_end;	/* Original (unsplit) end. */
	unsigned long maddr[0];	/* Array of MSPEC addresses. */
};

#define VMD_VMALLOCED 0x1	/* vmalloc'd rather than kmalloc'd */

/* used on shub2 to clear FOP cache in the HUB */
static unsigned long scratch_page[MAX_NUMNODES];
#define SH2_AMO_CACHE_ENTRIES	4

static inline int
mspec_zero_block(unsigned long addr, int len)
{
	int status;

	if (is_sn2) {
		if (is_shub2()) {
			int nid;
			void *p;
			int i;

			nid = nasid_to_cnodeid(get_node_number(__pa(addr)));
			p = (void *)TO_AMO(scratch_page[nid]);

			for (i=0; i < SH2_AMO_CACHE_ENTRIES; i++) {
				FETCHOP_LOAD_OP(p, FETCHOP_LOAD);
				p += FETCHOP_VAR_SIZE;
			}
		}

		status = bte_copy(0, addr & ~__IA64_UNCACHED_OFFSET, len,
				  BTE_WACQUIRE | BTE_ZERO_FILL, NULL);
	} else {
		memset((char *) addr, 0, len);
		status = 0;
	}
	return status;
}

/*
 * mspec_open
 *
 * Called when a device mapping is created by a means other than mmap
 * (via fork, munmap, etc.).  Increments the reference count on the
 * underlying mspec data so it is not freed prematurely.
 */
static void
mspec_open(struct vm_area_struct *vma)
{
	struct vma_data *vdata;

	vdata = vma->vm_private_data;
	atomic_inc(&vdata->refcnt);
}

/*
 * mspec_close
 *
 * Called when unmapping a device mapping. Frees all mspec pages
 * belonging to all the vma's sharing this vma_data structure.
 */
static void
mspec_close(struct vm_area_struct *vma)
{
	struct vma_data *vdata;
	int index, last_index;
	unsigned long my_page;

	vdata = vma->vm_private_data;

	if (!atomic_dec_and_test(&vdata->refcnt))
		return;

	last_index = (vdata->vm_end - vdata->vm_start) >> PAGE_SHIFT;
	for (index = 0; index < last_index; index++) {
		if (vdata->maddr[index] == 0)
			continue;
		/*
		 * Clear the page before sticking it back
		 * into the pool.
		 */
		my_page = vdata->maddr[index];
		vdata->maddr[index] = 0;
		if (!mspec_zero_block(my_page, PAGE_SIZE))
			uncached_free_page(my_page, 1);
		else
			printk(KERN_WARNING "mspec_close(): "
			       "failed to zero page %ld\n", my_page);
	}

	if (vdata->flags & VMD_VMALLOCED)
		vfree(vdata);
	else
		kfree(vdata);
}

/*
 * mspec_fault
 *
 * Creates a mspec page and maps it to user space.
 */
static int
mspec_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
	unsigned long paddr, maddr;
	unsigned long pfn;
	pgoff_t index = vmf->pgoff;
	struct vma_data *vdata = vma->vm_private_data;

	maddr = (volatile unsigned long) vdata->maddr[index];
	if (maddr == 0) {
		maddr = uncached_alloc_page(numa_node_id(), 1);
		if (maddr == 0)
			return VM_FAULT_OOM;

		spin_lock(&vdata->lock);
		if (vdata->maddr[index] == 0) {
			vdata->count++;
			vdata->maddr[index] = maddr;
		} else {
			uncached_free_page(maddr, 1);
			maddr = vdata->maddr[index];
		}
		spin_unlock(&vdata->lock);
	}

	if (vdata->type == MSPEC_FETCHOP)
		paddr = TO_AMO(maddr);
	else
		paddr = maddr & ~__IA64_UNCACHED_OFFSET;

	pfn = paddr >> PAGE_SHIFT;

	/*
	 * vm_insert_pfn can fail with -EBUSY, but in that case it will
	 * be because another thread has installed the pte first, so it
	 * is no problem.
	 */
	vm_insert_pfn(vma, (unsigned long)vmf->virtual_address, pfn);

	return VM_FAULT_NOPAGE;
}

static const struct vm_operations_struct mspec_vm_ops = {
	.open = mspec_open,
	.close = mspec_close,
	.fault = mspec_fault,
};

/*
 * mspec_mmap
 *
 * Called when mmapping the device.  Initializes the vma with a fault handler
 * and private data structure necessary to allocate, track, and free the
 * underlying pages.
 */
static int
mspec_mmap(struct file *file, struct vm_area_struct *vma,
					enum mspec_page_type type)
{
	struct vma_data *vdata;
	int pages, vdata_size, flags = 0;

	if (vma->vm_pgoff != 0)
		return -EINVAL;

	if ((vma->vm_flags & VM_SHARED) == 0)
		return -EINVAL;

	if ((vma->vm_flags & VM_WRITE) == 0)
		return -EPERM;

	pages = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
	vdata_size = sizeof(struct vma_data) + pages * sizeof(long);
	if (vdata_size <= PAGE_SIZE)
		vdata = kmalloc(vdata_size, GFP_KERNEL);
	else {
		vdata = vmalloc(vdata_size);
		flags = VMD_VMALLOCED;
	}
	if (!vdata)
		return -ENOMEM;
	memset(vdata, 0, vdata_size);

	vdata->vm_start = vma->vm_start;
	vdata->vm_end = vma->vm_end;
	vdata->flags = flags;
	vdata->type = type;
	spin_lock_init(&vdata->lock);
	vdata->refcnt = ATOMIC_INIT(1);
	vma->vm_private_data = vdata;

	vma->vm_flags |= (VM_IO | VM_RESERVED | VM_PFNMAP | VM_DONTEXPAND);
	if (vdata->type == MSPEC_FETCHOP || vdata->type == MSPEC_UNCACHED)
		vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
	vma->vm_ops = &mspec_vm_ops;

	return 0;
}

static int
fetchop_mmap(struct file *file, struct vm_area_struct *vma)
{
	return mspec_mmap(file, vma, MSPEC_FETCHOP);
}

static int
cached_mmap(struct file *file, struct vm_area_struct *vma)
{
	return mspec_mmap(file, vma, MSPEC_CACHED);
}

static int
uncached_mmap(struct file *file, struct vm_area_struct *vma)
{
	return mspec_mmap(file, vma, MSPEC_UNCACHED);
}

static const struct file_operations fetchop_fops = {
	.owner = THIS_MODULE,
	.mmap = fetchop_mmap,
	.llseek = noop_llseek,
};

static struct miscdevice fetchop_miscdev = {
	.minor = MISC_DYNAMIC_MINOR,
	.name = "sgi_fetchop",
	.fops = &fetchop_fops
};

static const struct file_operations cached_fops = {
	.owner = THIS_MODULE,
	.mmap = cached_mmap,
	.llseek = noop_llseek,
};

static struct miscdevice cached_miscdev = {
	.minor = MISC_DYNAMIC_MINOR,
	.name = "mspec_cached",
	.fops = &cached_fops
};

static const struct file_operations uncached_fops = {
	.owner = THIS_MODULE,
	.mmap = uncached_mmap,
	.llseek = noop_llseek,
};

static struct miscdevice uncached_miscdev = {
	.minor = MISC_DYNAMIC_MINOR,
	.name = "mspec_uncached",
	.fops = &uncached_fops
};

/*
 * mspec_init
 *
 * Called at boot time to initialize the mspec facility.
 */
static int __init
mspec_init(void)
{
	int ret;
	int nid;

	/*
	 * The fetchop device only works on SN2 hardware, uncached and cached
	 * memory drivers should both be valid on all ia64 hardware
	 */
#ifdef CONFIG_SGI_SN
	if (ia64_platform_is("sn2")) {
		is_sn2 = 1;
		if (is_shub2()) {
			ret = -ENOMEM;
			for_each_node_state(nid, N_ONLINE) {
				int actual_nid;
				int nasid;
				unsigned long phys;

				scratch_page[nid] = uncached_alloc_page(nid, 1);
				if (scratch_page[nid] == 0)
					goto free_scratch_pages;
				phys = __pa(scratch_page[nid]);
				nasid = get_node_number(phys);
				actual_nid = nasid_to_cnodeid(nasid);
				if (actual_nid != nid)
					goto free_scratch_pages;
			}
		}

		ret = misc_register(&fetchop_miscdev);
		if (ret) {
			printk(KERN_ERR
			       "%s: failed to register device %i\n",
			       FETCHOP_ID, ret);
			goto free_scratch_pages;
		}
	}
#endif
	ret = misc_register(&cached_miscdev);
	if (ret) {
		printk(KERN_ERR "%s: failed to register device %i\n",
		       CACHED_ID, ret);
		if (is_sn2)
			misc_deregister(&fetchop_miscdev);
		goto free_scratch_pages;
	}
	ret = misc_register(&uncached_miscdev);
	if (ret) {
		printk(KERN_ERR "%s: failed to register device %i\n",
		       UNCACHED_ID, ret);
		misc_deregister(&cached_miscdev);
		if (is_sn2)
			misc_deregister(&fetchop_miscdev);
		goto free_scratch_pages;
	}

	printk(KERN_INFO "%s %s initialized devices: %s %s %s\n",
	       MSPEC_BASENAME, REVISION, is_sn2 ? FETCHOP_ID : "",
	       CACHED_ID, UNCACHED_ID);

	return 0;

 free_scratch_pages:
	for_each_node(nid) {
		if (scratch_page[nid] != 0)
			uncached_free_page(scratch_page[nid], 1);
	}
	return ret;
}

static void __exit
mspec_exit(void)
{
	int nid;

	misc_deregister(&uncached_miscdev);
	misc_deregister(&cached_miscdev);
	if (is_sn2) {
		misc_deregister(&fetchop_miscdev);

		for_each_node(nid) {
			if (scratch_page[nid] != 0)
				uncached_free_page(scratch_page[nid], 1);
		}
	}
}

module_init(mspec_init);
module_exit(mspec_exit);

MODULE_AUTHOR("Silicon Graphics, Inc. <linux-altix@sgi.com>");
MODULE_DESCRIPTION("Driver for SGI SN special memory operations");
MODULE_LICENSE("GPL");
Commit	Line	Data
17a3b050 JS	1	/*
	2	* Copyright (C) 2001-2006 Silicon Graphics, Inc. All rights
	3	* reserved.
	4	*
	5	* This program is free software; you can redistribute it and/or modify it
	6	* under the terms of version 2 of the GNU General Public License
	7	* as published by the Free Software Foundation.
	8	*/
	9
	10	/*
	11	* SN Platform Special Memory (mspec) Support
	12	*
	13	* This driver exports the SN special memory (mspec) facility to user
	14	* processes.
	15	* There are three types of memory made available thru this driver:
	16	* fetchops, uncached and cached.
	17	*
	18	* Fetchops are atomic memory operations that are implemented in the
	19	* memory controller on SGI SN hardware.
	20	*
	21	* Uncached are used for memory write combining feature of the ia64
	22	* cpu.
	23	*
	24	* Cached are used for areas of memory that are used as cached addresses
	25	* on our partition and used as uncached addresses from other partitions.
	26	* Due to a design constraint of the SN2 Shub, you can not have processors
	27	* on the same FSB perform both a cached and uncached reference to the
	28	* same cache line. These special memory cached regions prevent the
	29	* kernel from ever dropping in a TLB entry and therefore prevent the
	30	* processor from ever speculating a cache line from this page.
	31	*/
	32
17a3b050 JS	33	#include <linux/types.h>
	34	#include <linux/kernel.h>
	35	#include <linux/module.h>
	36	#include <linux/init.h>
	37	#include <linux/errno.h>
	38	#include <linux/miscdevice.h>
	39	#include <linux/spinlock.h>
	40	#include <linux/mm.h>
4e950f6f	41	#include <linux/fs.h>
17a3b050 JS	42	#include <linux/vmalloc.h>
	43	#include <linux/string.h>
	44	#include <linux/slab.h>
	45	#include <linux/numa.h>
	46	#include <asm/page.h>
	47	#include <asm/system.h>
	48	#include <asm/pgtable.h>
	49	#include <asm/atomic.h>
	50	#include <asm/tlbflush.h>
	51	#include <asm/uncached.h>
	52	#include <asm/sn/addrs.h>
	53	#include <asm/sn/arch.h>
	54	#include <asm/sn/mspec.h>
	55	#include <asm/sn/sn_cpuid.h>
	56	#include <asm/sn/io.h>
	57	#include <asm/sn/bte.h>
	58	#include <asm/sn/shubio.h>
	59
	60
	61	#define FETCHOP_ID "SGI Fetchop,"
	62	#define CACHED_ID "Cached,"
	63	#define UNCACHED_ID "Uncached"
	64	#define REVISION "4.0"
	65	#define MSPEC_BASENAME "mspec"
	66
	67	/*
	68	* Page types allocated by the device.
	69	*/
4191ba26	70	enum mspec_page_type {
17a3b050 JS	71	MSPEC_FETCHOP = 1,
	72	MSPEC_CACHED,
	73	MSPEC_UNCACHED
	74	};
	75
1a4b0fc5	76	#ifdef CONFIG_SGI_SN
17a3b050	77	static int is_sn2;
1a4b0fc5 JS	78	#else
	79	#define is_sn2 0
	80	#endif
17a3b050 JS	81
	82	/*
	83	* One of these structures is allocated when an mspec region is mmaped. The
	84	* structure is pointed to by the vma->vm_private_data field in the vma struct.
	85	* This structure is used to record the addresses of the mspec pages.
4191ba26 CW	86	* This structure is shared by all vma's that are split off from the
	87	* original vma when split_vma()'s are done.
	88	*
	89	* The refcnt is incremented atomically because mm->mmap_sem does not
	90	* protect in fork case where multiple tasks share the vma_data.
17a3b050 JS	91	*/
	92	struct vma_data {
	93	atomic_t refcnt; /* Number of vmas sharing the data. */
4191ba26	94	spinlock_t lock; /* Serialize access to this structure. */
17a3b050	95	int count; /* Number of pages allocated. */
4191ba26 CW	96	enum mspec_page_type type; /* Type of pages allocated. */
	97	int flags; /* See VMD_xxx below. */
	98	unsigned long vm_start; /* Original (unsplit) base. */
	99	unsigned long vm_end; /* Original (unsplit) end. */
17a3b050 JS	100	unsigned long maddr[0]; /* Array of MSPEC addresses. */
	101	};
	102
4191ba26 CW	103	#define VMD_VMALLOCED 0x1 /* vmalloc'd rather than kmalloc'd */
4191ba26 CW	104
17a3b050 JS	105	/* used on shub2 to clear FOP cache in the HUB */
	106	static unsigned long scratch_page[MAX_NUMNODES];
	107	#define SH2_AMO_CACHE_ENTRIES 4
	108
	109	static inline int
	110	mspec_zero_block(unsigned long addr, int len)
	111	{
	112	int status;
	113
	114	if (is_sn2) {
	115	if (is_shub2()) {
	116	int nid;
	117	void *p;
	118	int i;
	119
	120	nid = nasid_to_cnodeid(get_node_number(__pa(addr)));
	121	p = (void *)TO_AMO(scratch_page[nid]);
	122
	123	for (i=0; i < SH2_AMO_CACHE_ENTRIES; i++) {
	124	FETCHOP_LOAD_OP(p, FETCHOP_LOAD);
	125	p += FETCHOP_VAR_SIZE;
	126	}
	127	}
	128
	129	status = bte_copy(0, addr & ~__IA64_UNCACHED_OFFSET, len,
	130	BTE_WACQUIRE \| BTE_ZERO_FILL, NULL);
	131	} else {
	132	memset((char *) addr, 0, len);
	133	status = 0;
	134	}
	135	return status;
	136	}
	137
	138	/*
	139	* mspec_open
	140	*
	141	* Called when a device mapping is created by a means other than mmap
4191ba26 CW	142	* (via fork, munmap, etc.). Increments the reference count on the
4191ba26 CW	143	* underlying mspec data so it is not freed prematurely.
17a3b050 JS	144	*/
	145	static void
	146	mspec_open(struct vm_area_struct *vma)
	147	{
	148	struct vma_data *vdata;
	149
	150	vdata = vma->vm_private_data;
	151	atomic_inc(&vdata->refcnt);
	152	}
	153
	154	/*
	155	* mspec_close
	156	*
	157	* Called when unmapping a device mapping. Frees all mspec pages
afa684f6	158	* belonging to all the vma's sharing this vma_data structure.
17a3b050 JS	159	*/
	160	static void
	161	mspec_close(struct vm_area_struct *vma)
	162	{
	163	struct vma_data *vdata;
afa684f6	164	int index, last_index;
4191ba26	165	unsigned long my_page;
17a3b050 JS	166
17a3b050 JS	167	vdata = vma->vm_private_data;
17a3b050	168
afa684f6 CW	169	if (!atomic_dec_and_test(&vdata->refcnt))
afa684f6 CW	170	return;
4191ba26	171
afa684f6 CW	172	last_index = (vdata->vm_end - vdata->vm_start) >> PAGE_SHIFT;
afa684f6 CW	173	for (index = 0; index < last_index; index++) {
4191ba26	174	if (vdata->maddr[index] == 0)
17a3b050 JS	175	continue;
	176	/*
	177	* Clear the page before sticking it back
	178	* into the pool.
	179	*/
4191ba26 CW	180	my_page = vdata->maddr[index];
4191ba26 CW	181	vdata->maddr[index] = 0;
afa684f6	182	if (!mspec_zero_block(my_page, PAGE_SIZE))
e4a064df	183	uncached_free_page(my_page, 1);
17a3b050 JS	184	else
17a3b050 JS	185	printk(KERN_WARNING "mspec_close(): "
afa684f6	186	"failed to zero page %ld\n", my_page);
17a3b050	187	}
4191ba26 CW	188
4191ba26 CW	189	if (vdata->flags & VMD_VMALLOCED)
17a3b050	190	vfree(vdata);
4191ba26 CW	191	else
4191ba26 CW	192	kfree(vdata);
17a3b050 JS	193	}
17a3b050 JS	194
17a3b050	195	/*
efe9e779	196	* mspec_fault
17a3b050 JS	197	*
	198	* Creates a mspec page and maps it to user space.
	199	*/
efe9e779 NP	200	static int
efe9e779 NP	201	mspec_fault(struct vm_area_struct vma, struct vm_fault vmf)
17a3b050 JS	202	{
	203	unsigned long paddr, maddr;
	204	unsigned long pfn;
efe9e779	205	pgoff_t index = vmf->pgoff;
17a3b050 JS	206	struct vma_data *vdata = vma->vm_private_data;
17a3b050 JS	207
17a3b050 JS	208	maddr = (volatile unsigned long) vdata->maddr[index];
17a3b050 JS	209	if (maddr == 0) {
e4a064df	210	maddr = uncached_alloc_page(numa_node_id(), 1);
17a3b050	211	if (maddr == 0)
efe9e779	212	return VM_FAULT_OOM;
17a3b050 JS	213
	214	spin_lock(&vdata->lock);
	215	if (vdata->maddr[index] == 0) {
	216	vdata->count++;
	217	vdata->maddr[index] = maddr;
	218	} else {
e4a064df	219	uncached_free_page(maddr, 1);
17a3b050 JS	220	maddr = vdata->maddr[index];
	221	}
	222	spin_unlock(&vdata->lock);
	223	}
	224
	225	if (vdata->type == MSPEC_FETCHOP)
	226	paddr = TO_AMO(maddr);
	227	else
1a4b0fc5	228	paddr = maddr & ~__IA64_UNCACHED_OFFSET;
17a3b050 JS	229
	230	pfn = paddr >> PAGE_SHIFT;
	231
efe9e779 NP	232	/*
	233	* vm_insert_pfn can fail with -EBUSY, but in that case it will
	234	* be because another thread has installed the pte first, so it
	235	* is no problem.
	236	*/
	237	vm_insert_pfn(vma, (unsigned long)vmf->virtual_address, pfn);
	238
	239	return VM_FAULT_NOPAGE;
17a3b050 JS	240	}
17a3b050 JS	241
f0f37e2f	242	static const struct vm_operations_struct mspec_vm_ops = {
17a3b050 JS	243	.open = mspec_open,
17a3b050 JS	244	.close = mspec_close,
efe9e779	245	.fault = mspec_fault,
17a3b050 JS	246	};
	247
	248	/*
	249	* mspec_mmap
	250	*
af901ca1	251	* Called when mmapping the device. Initializes the vma with a fault handler
17a3b050 JS	252	* and private data structure necessary to allocate, track, and free the
	253	* underlying pages.
	254	*/
	255	static int
4191ba26 CW	256	mspec_mmap(struct file file, struct vm_area_struct vma,
4191ba26 CW	257	enum mspec_page_type type)
17a3b050 JS	258	{
17a3b050 JS	259	struct vma_data *vdata;
4191ba26	260	int pages, vdata_size, flags = 0;
17a3b050 JS	261
	262	if (vma->vm_pgoff != 0)
	263	return -EINVAL;
	264
	265	if ((vma->vm_flags & VM_SHARED) == 0)
	266	return -EINVAL;
	267
	268	if ((vma->vm_flags & VM_WRITE) == 0)
	269	return -EPERM;
	270
	271	pages = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
	272	vdata_size = sizeof(struct vma_data) + pages * sizeof(long);
	273	if (vdata_size <= PAGE_SIZE)
	274	vdata = kmalloc(vdata_size, GFP_KERNEL);
4191ba26	275	else {
17a3b050	276	vdata = vmalloc(vdata_size);
4191ba26 CW	277	flags = VMD_VMALLOCED;
4191ba26 CW	278	}
17a3b050 JS	279	if (!vdata)
	280	return -ENOMEM;
	281	memset(vdata, 0, vdata_size);
	282
4191ba26 CW	283	vdata->vm_start = vma->vm_start;
	284	vdata->vm_end = vma->vm_end;
	285	vdata->flags = flags;
17a3b050 JS	286	vdata->type = type;
	287	spin_lock_init(&vdata->lock);
	288	vdata->refcnt = ATOMIC_INIT(1);
	289	vma->vm_private_data = vdata;
	290
2f98735c	291	vma->vm_flags \|= (VM_IO \| VM_RESERVED \| VM_PFNMAP \| VM_DONTEXPAND);
17a3b050 JS	292	if (vdata->type == MSPEC_FETCHOP \|\| vdata->type == MSPEC_UNCACHED)
	293	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
	294	vma->vm_ops = &mspec_vm_ops;
	295
	296	return 0;
	297	}
	298
	299	static int
	300	fetchop_mmap(struct file file, struct vm_area_struct vma)
	301	{
	302	return mspec_mmap(file, vma, MSPEC_FETCHOP);
	303	}
	304
	305	static int
	306	cached_mmap(struct file file, struct vm_area_struct vma)
	307	{
	308	return mspec_mmap(file, vma, MSPEC_CACHED);
	309	}
	310
	311	static int
	312	uncached_mmap(struct file file, struct vm_area_struct vma)
	313	{
	314	return mspec_mmap(file, vma, MSPEC_UNCACHED);
	315	}
	316
2b8693c0	317	static const struct file_operations fetchop_fops = {
17a3b050	318	.owner = THIS_MODULE,
6038f373 AB	319	.mmap = fetchop_mmap,
6038f373 AB	320	.llseek = noop_llseek,
17a3b050 JS	321	};
	322
	323	static struct miscdevice fetchop_miscdev = {
	324	.minor = MISC_DYNAMIC_MINOR,
	325	.name = "sgi_fetchop",
	326	.fops = &fetchop_fops
	327	};
	328
2b8693c0	329	static const struct file_operations cached_fops = {
17a3b050	330	.owner = THIS_MODULE,
6038f373 AB	331	.mmap = cached_mmap,
6038f373 AB	332	.llseek = noop_llseek,
17a3b050 JS	333	};
	334
	335	static struct miscdevice cached_miscdev = {
	336	.minor = MISC_DYNAMIC_MINOR,
	337	.name = "mspec_cached",
	338	.fops = &cached_fops
	339	};
	340
2b8693c0	341	static const struct file_operations uncached_fops = {
17a3b050	342	.owner = THIS_MODULE,
6038f373 AB	343	.mmap = uncached_mmap,
6038f373 AB	344	.llseek = noop_llseek,
17a3b050 JS	345	};
	346
	347	static struct miscdevice uncached_miscdev = {
	348	.minor = MISC_DYNAMIC_MINOR,
	349	.name = "mspec_uncached",
	350	.fops = &uncached_fops
	351	};
	352
	353	/*
	354	* mspec_init
	355	*
	356	* Called at boot time to initialize the mspec facility.
	357	*/
	358	static int __init
	359	mspec_init(void)
	360	{
	361	int ret;
	362	int nid;
	363
	364	/*
	365	* The fetchop device only works on SN2 hardware, uncached and cached
	366	* memory drivers should both be valid on all ia64 hardware
	367	*/
1a4b0fc5	368	#ifdef CONFIG_SGI_SN
17a3b050 JS	369	if (ia64_platform_is("sn2")) {
	370	is_sn2 = 1;
	371	if (is_shub2()) {
	372	ret = -ENOMEM;
2dca53a9	373	for_each_node_state(nid, N_ONLINE) {
17a3b050 JS	374	int actual_nid;
	375	int nasid;
	376	unsigned long phys;
	377
e4a064df	378	scratch_page[nid] = uncached_alloc_page(nid, 1);
17a3b050 JS	379	if (scratch_page[nid] == 0)
	380	goto free_scratch_pages;
	381	phys = __pa(scratch_page[nid]);
	382	nasid = get_node_number(phys);
	383	actual_nid = nasid_to_cnodeid(nasid);
	384	if (actual_nid != nid)
	385	goto free_scratch_pages;
	386	}
	387	}
	388
	389	ret = misc_register(&fetchop_miscdev);
	390	if (ret) {
	391	printk(KERN_ERR
	392	"%s: failed to register device %i\n",
	393	FETCHOP_ID, ret);
	394	goto free_scratch_pages;
	395	}
	396	}
1a4b0fc5	397	#endif
17a3b050 JS	398	ret = misc_register(&cached_miscdev);
	399	if (ret) {
	400	printk(KERN_ERR "%s: failed to register device %i\n",
	401	CACHED_ID, ret);
	402	if (is_sn2)
	403	misc_deregister(&fetchop_miscdev);
	404	goto free_scratch_pages;
	405	}
	406	ret = misc_register(&uncached_miscdev);
	407	if (ret) {
	408	printk(KERN_ERR "%s: failed to register device %i\n",
	409	UNCACHED_ID, ret);
	410	misc_deregister(&cached_miscdev);
	411	if (is_sn2)
	412	misc_deregister(&fetchop_miscdev);
	413	goto free_scratch_pages;
	414	}
	415
	416	printk(KERN_INFO "%s %s initialized devices: %s %s %s\n",
	417	MSPEC_BASENAME, REVISION, is_sn2 ? FETCHOP_ID : "",
	418	CACHED_ID, UNCACHED_ID);
	419
	420	return 0;
	421
	422	free_scratch_pages:
	423	for_each_node(nid) {
	424	if (scratch_page[nid] != 0)
e4a064df	425	uncached_free_page(scratch_page[nid], 1);
17a3b050 JS	426	}
	427	return ret;
	428	}
	429
	430	static void __exit
	431	mspec_exit(void)
	432	{
	433	int nid;
	434
	435	misc_deregister(&uncached_miscdev);
	436	misc_deregister(&cached_miscdev);
	437	if (is_sn2) {
	438	misc_deregister(&fetchop_miscdev);
	439
	440	for_each_node(nid) {
	441	if (scratch_page[nid] != 0)
e4a064df	442	uncached_free_page(scratch_page[nid], 1);
17a3b050 JS	443	}
	444	}
	445	}
	446
	447	module_init(mspec_init);
	448	module_exit(mspec_exit);
	449
	450	MODULE_AUTHOR("Silicon Graphics, Inc. <linux-altix@sgi.com>");
	451	MODULE_DESCRIPTION("Driver for SGI SN special memory operations");
	452	MODULE_LICENSE("GPL");