1 /*-
2  * Copyright (c) 2010 Isilon Systems, Inc.
3  * Copyright (c) 2010 iX Systems, Inc.
4  * Copyright (c) 2010 Panasas, Inc.
5  * Copyright (c) 2013-2017 Mellanox Technologies, Ltd.
6  * Copyright (c) 2015 François Tigeot
7  * Copyright (c) 2015 Matthew Dillon <dillon@backplane.com>
8  * All rights reserved.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice unmodified, this list of conditions, and the following
15  *    disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
21  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
22  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
23  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
24  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
25  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
29  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30  */
31 #ifndef	_LINUXKPI_LINUX_MM_H_
32 #define	_LINUXKPI_LINUX_MM_H_
33 
34 #include <linux/spinlock.h>
35 #include <linux/gfp.h>
36 #include <linux/kernel.h>
37 #include <linux/mm_types.h>
38 #include <linux/pfn.h>
39 #include <linux/list.h>
40 #include <linux/mmap_lock.h>
41 #include <linux/overflow.h>
42 #include <linux/shrinker.h>
43 #include <linux/page.h>
44 
45 #include <asm/pgtable.h>
46 
47 #define	PAGE_ALIGN(x)	ALIGN(x, PAGE_SIZE)
48 
49 /*
50  * Make sure our LinuxKPI defined virtual memory flags don't conflict
51  * with the ones defined by FreeBSD:
52  */
53 CTASSERT((VM_PROT_ALL & -(1 << 8)) == 0);
54 
55 #define	VM_READ			VM_PROT_READ
56 #define	VM_WRITE		VM_PROT_WRITE
57 #define	VM_EXEC			VM_PROT_EXECUTE
58 
59 #define	VM_ACCESS_FLAGS		(VM_READ | VM_WRITE | VM_EXEC)
60 
61 #define	VM_PFNINTERNAL		(1 << 8)	/* FreeBSD private flag to vm_insert_pfn() */
62 #define	VM_MIXEDMAP		(1 << 9)
63 #define	VM_NORESERVE		(1 << 10)
64 #define	VM_PFNMAP		(1 << 11)
65 #define	VM_IO			(1 << 12)
66 #define	VM_MAYWRITE		(1 << 13)
67 #define	VM_DONTCOPY		(1 << 14)
68 #define	VM_DONTEXPAND		(1 << 15)
69 #define	VM_DONTDUMP		(1 << 16)
70 #define	VM_SHARED		(1 << 17)
71 
72 #define	VMA_MAX_PREFAULT_RECORD	1
73 
74 #define	FOLL_WRITE		(1 << 0)
75 #define	FOLL_FORCE		(1 << 1)
76 
77 #define	VM_FAULT_OOM		(1 << 0)
78 #define	VM_FAULT_SIGBUS		(1 << 1)
79 #define	VM_FAULT_MAJOR		(1 << 2)
80 #define	VM_FAULT_WRITE		(1 << 3)
81 #define	VM_FAULT_HWPOISON	(1 << 4)
82 #define	VM_FAULT_HWPOISON_LARGE	(1 << 5)
83 #define	VM_FAULT_SIGSEGV	(1 << 6)
84 #define	VM_FAULT_NOPAGE		(1 << 7)
85 #define	VM_FAULT_LOCKED		(1 << 8)
86 #define	VM_FAULT_RETRY		(1 << 9)
87 #define	VM_FAULT_FALLBACK	(1 << 10)
88 
89 #define	VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
90 	VM_FAULT_HWPOISON |VM_FAULT_HWPOISON_LARGE | VM_FAULT_FALLBACK)
91 
92 #define	FAULT_FLAG_WRITE	(1 << 0)
93 #define	FAULT_FLAG_MKWRITE	(1 << 1)
94 #define	FAULT_FLAG_ALLOW_RETRY	(1 << 2)
95 #define	FAULT_FLAG_RETRY_NOWAIT	(1 << 3)
96 #define	FAULT_FLAG_KILLABLE	(1 << 4)
97 #define	FAULT_FLAG_TRIED	(1 << 5)
98 #define	FAULT_FLAG_USER		(1 << 6)
99 #define	FAULT_FLAG_REMOTE	(1 << 7)
100 #define	FAULT_FLAG_INSTRUCTION	(1 << 8)
101 
102 #define fault_flag_allow_retry_first(flags) \
103 	(((flags) & (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_TRIED)) == FAULT_FLAG_ALLOW_RETRY)
104 
105 typedef int (*pte_fn_t)(linux_pte_t *, unsigned long addr, void *data);
106 
107 struct vm_area_struct {
108 	vm_offset_t vm_start;
109 	vm_offset_t vm_end;
110 	vm_offset_t vm_pgoff;
111 	pgprot_t vm_page_prot;
112 	unsigned long vm_flags;
113 	struct mm_struct *vm_mm;
114 	void   *vm_private_data;
115 	const struct vm_operations_struct *vm_ops;
116 	struct linux_file *vm_file;
117 
118 	/* internal operation */
119 	vm_paddr_t vm_pfn;		/* PFN for memory map */
120 	vm_size_t vm_len;		/* length for memory map */
121 	vm_pindex_t vm_pfn_first;
122 	int	vm_pfn_count;
123 	int    *vm_pfn_pcount;
124 	vm_object_t vm_obj;
125 	vm_map_t vm_cached_map;
126 	TAILQ_ENTRY(vm_area_struct) vm_entry;
127 };
128 
129 struct vm_fault {
130 	unsigned int flags;
131 	pgoff_t	pgoff;
132 	union {
133 		/* user-space address */
134 		void *virtual_address;	/* < 4.11 */
135 		unsigned long address;	/* >= 4.11 */
136 	};
137 	struct page *page;
138 	struct vm_area_struct *vma;
139 };
140 
141 struct vm_operations_struct {
142 	void    (*open) (struct vm_area_struct *);
143 	void    (*close) (struct vm_area_struct *);
144 	int     (*fault) (struct vm_fault *);
145 	int	(*access) (struct vm_area_struct *, unsigned long, void *, int, int);
146 };
147 
148 struct sysinfo {
149 	uint64_t totalram;	/* Total usable main memory size */
150 	uint64_t freeram;	/* Available memory size */
151 	uint64_t totalhigh;	/* Total high memory size */
152 	uint64_t freehigh;	/* Available high memory size */
153 	uint32_t mem_unit;	/* Memory unit size in bytes */
154 };
155 
156 static inline struct page *
virt_to_head_page(const void * p)157 virt_to_head_page(const void *p)
158 {
159 
160 	return (virt_to_page(p));
161 }
162 
163 /*
164  * Compute log2 of the power of two rounded up count of pages
165  * needed for size bytes.
166  */
167 static inline int
get_order(unsigned long size)168 get_order(unsigned long size)
169 {
170 	int order;
171 
172 	size = (size - 1) >> PAGE_SHIFT;
173 	order = 0;
174 	while (size) {
175 		order++;
176 		size >>= 1;
177 	}
178 	return (order);
179 }
180 
181 /*
182  * Resolve a page into a virtual address:
183  *
184  * NOTE: This function only works for pages allocated by the kernel.
185  */
186 void *linux_page_address(struct page *);
187 #define	page_address(page) linux_page_address(page)
188 
189 static inline void *
lowmem_page_address(struct page * page)190 lowmem_page_address(struct page *page)
191 {
192 	return (page_address(page));
193 }
194 
195 /*
196  * This only works via memory map operations.
197  */
198 static inline int
io_remap_pfn_range(struct vm_area_struct * vma,unsigned long addr,unsigned long pfn,unsigned long size,vm_memattr_t prot)199 io_remap_pfn_range(struct vm_area_struct *vma,
200     unsigned long addr, unsigned long pfn, unsigned long size,
201     vm_memattr_t prot)
202 {
203 	vma->vm_page_prot = prot;
204 	vma->vm_pfn = pfn;
205 	vma->vm_len = size;
206 
207 	return (0);
208 }
209 
210 vm_fault_t
211 lkpi_vmf_insert_pfn_prot_locked(struct vm_area_struct *vma, unsigned long addr,
212     unsigned long pfn, pgprot_t prot);
213 
214 static inline vm_fault_t
vmf_insert_pfn_prot(struct vm_area_struct * vma,unsigned long addr,unsigned long pfn,pgprot_t prot)215 vmf_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
216     unsigned long pfn, pgprot_t prot)
217 {
218 	vm_fault_t ret;
219 
220 	VM_OBJECT_WLOCK(vma->vm_obj);
221 	ret = lkpi_vmf_insert_pfn_prot_locked(vma, addr, pfn, prot);
222 	VM_OBJECT_WUNLOCK(vma->vm_obj);
223 
224 	return (ret);
225 }
226 #define	vmf_insert_pfn_prot(...)	\
227 	_Static_assert(false,		\
228 "This function is always called in a loop. Consider using the locked version")
229 
230 static inline int
apply_to_page_range(struct mm_struct * mm,unsigned long address,unsigned long size,pte_fn_t fn,void * data)231 apply_to_page_range(struct mm_struct *mm, unsigned long address,
232     unsigned long size, pte_fn_t fn, void *data)
233 {
234 	return (-ENOTSUP);
235 }
236 
237 int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
238     unsigned long size);
239 
240 int lkpi_remap_pfn_range(struct vm_area_struct *vma,
241     unsigned long start_addr, unsigned long start_pfn, unsigned long size,
242     pgprot_t prot);
243 
244 static inline int
remap_pfn_range(struct vm_area_struct * vma,unsigned long addr,unsigned long pfn,unsigned long size,pgprot_t prot)245 remap_pfn_range(struct vm_area_struct *vma, unsigned long addr,
246     unsigned long pfn, unsigned long size, pgprot_t prot)
247 {
248 	return (lkpi_remap_pfn_range(vma, addr, pfn, size, prot));
249 }
250 
251 static inline unsigned long
vma_pages(struct vm_area_struct * vma)252 vma_pages(struct vm_area_struct *vma)
253 {
254 	return ((vma->vm_end - vma->vm_start) >> PAGE_SHIFT);
255 }
256 
257 #define	offset_in_page(off)	((unsigned long)(off) & (PAGE_SIZE - 1))
258 
259 static inline void
set_page_dirty(struct page * page)260 set_page_dirty(struct page *page)
261 {
262 	vm_page_dirty(page);
263 }
264 
265 static inline void
mark_page_accessed(struct page * page)266 mark_page_accessed(struct page *page)
267 {
268 	vm_page_reference(page);
269 }
270 
271 static inline void
get_page(struct page * page)272 get_page(struct page *page)
273 {
274 	vm_page_wire(page);
275 }
276 
277 extern long
278 get_user_pages(unsigned long start, unsigned long nr_pages,
279     unsigned int gup_flags, struct page **,
280     struct vm_area_struct **);
281 
282 static inline long
pin_user_pages(unsigned long start,unsigned long nr_pages,unsigned int gup_flags,struct page ** pages,struct vm_area_struct ** vmas)283 pin_user_pages(unsigned long start, unsigned long nr_pages,
284     unsigned int gup_flags, struct page **pages,
285     struct vm_area_struct **vmas)
286 {
287 	return get_user_pages(start, nr_pages, gup_flags, pages, vmas);
288 }
289 
290 extern int
291 __get_user_pages_fast(unsigned long start, int nr_pages, int write,
292     struct page **);
293 
294 static inline int
pin_user_pages_fast(unsigned long start,int nr_pages,unsigned int gup_flags,struct page ** pages)295 pin_user_pages_fast(unsigned long start, int nr_pages,
296     unsigned int gup_flags, struct page **pages)
297 {
298 	return __get_user_pages_fast(
299 	    start, nr_pages, !!(gup_flags & FOLL_WRITE), pages);
300 }
301 
302 extern long
303 get_user_pages_remote(struct task_struct *, struct mm_struct *,
304     unsigned long start, unsigned long nr_pages,
305     unsigned int gup_flags, struct page **,
306     struct vm_area_struct **);
307 
308 static inline long
pin_user_pages_remote(struct task_struct * task,struct mm_struct * mm,unsigned long start,unsigned long nr_pages,unsigned int gup_flags,struct page ** pages,struct vm_area_struct ** vmas)309 pin_user_pages_remote(struct task_struct *task, struct mm_struct *mm,
310     unsigned long start, unsigned long nr_pages,
311     unsigned int gup_flags, struct page **pages,
312     struct vm_area_struct **vmas)
313 {
314 	return get_user_pages_remote(
315 	    task, mm, start, nr_pages, gup_flags, pages, vmas);
316 }
317 
318 static inline void
put_page(struct page * page)319 put_page(struct page *page)
320 {
321 	vm_page_unwire(page, PQ_ACTIVE);
322 }
323 
324 #define	unpin_user_page(page) put_page(page)
325 #define	unpin_user_pages(pages, npages) release_pages(pages, npages)
326 
327 #define	copy_highpage(to, from) pmap_copy_page(from, to)
328 
329 static inline pgprot_t
vm_get_page_prot(unsigned long vm_flags)330 vm_get_page_prot(unsigned long vm_flags)
331 {
332 	return (vm_flags & VM_PROT_ALL);
333 }
334 
335 static inline void
vm_flags_set(struct vm_area_struct * vma,unsigned long flags)336 vm_flags_set(struct vm_area_struct *vma, unsigned long flags)
337 {
338 	vma->vm_flags |= flags;
339 }
340 
341 static inline void
vm_flags_clear(struct vm_area_struct * vma,unsigned long flags)342 vm_flags_clear(struct vm_area_struct *vma, unsigned long flags)
343 {
344 	vma->vm_flags &= ~flags;
345 }
346 
347 static inline struct page *
vmalloc_to_page(const void * addr)348 vmalloc_to_page(const void *addr)
349 {
350 	vm_paddr_t paddr;
351 
352 	paddr = pmap_kextract((vm_offset_t)addr);
353 	return (PHYS_TO_VM_PAGE(paddr));
354 }
355 
356 static inline int
trylock_page(struct page * page)357 trylock_page(struct page *page)
358 {
359 	return (vm_page_trylock(page));
360 }
361 
362 static inline void
unlock_page(struct page * page)363 unlock_page(struct page *page)
364 {
365 
366 	vm_page_unlock(page);
367 }
368 
369 extern int is_vmalloc_addr(const void *addr);
370 void si_meminfo(struct sysinfo *si);
371 
372 static inline unsigned long
totalram_pages(void)373 totalram_pages(void)
374 {
375 	return ((unsigned long)physmem);
376 }
377 
378 #define	unmap_mapping_range(...)	lkpi_unmap_mapping_range(__VA_ARGS__)
379 void lkpi_unmap_mapping_range(void *obj, loff_t const holebegin __unused,
380     loff_t const holelen, int even_cows __unused);
381 
382 #define PAGE_ALIGNED(p)	__is_aligned(p, PAGE_SIZE)
383 
384 void vma_set_file(struct vm_area_struct *vma, struct linux_file *file);
385 
386 static inline void
might_alloc(gfp_t gfp_mask __unused)387 might_alloc(gfp_t gfp_mask __unused)
388 {
389 }
390 
391 #define	is_cow_mapping(flags)	(false)
392 
393 static inline bool
want_init_on_free(void)394 want_init_on_free(void)
395 {
396 	return (false);
397 }
398 
399 #endif					/* _LINUXKPI_LINUX_MM_H_ */
400