2 * Copyright (c) 2010 Isilon Systems, Inc.
3 * Copyright (c) 2016 Matt Macy (mmacy@nextbsd.org)
4 * Copyright (c) 2017 Mellanox Technologies, Ltd.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice unmodified, this list of conditions, and the following
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
32 #include <sys/param.h>
33 #include <sys/systm.h>
34 #include <sys/malloc.h>
35 #include <sys/kernel.h>
36 #include <sys/sysctl.h>
38 #include <sys/mutex.h>
39 #include <sys/rwlock.h>
41 #include <sys/sched.h>
43 #include <machine/bus.h>
47 #include <vm/vm_param.h>
48 #include <vm/vm_kern.h>
49 #include <vm/vm_object.h>
50 #include <vm/vm_map.h>
51 #include <vm/vm_page.h>
52 #include <vm/vm_pageout.h>
53 #include <vm/vm_pager.h>
54 #include <vm/vm_radix.h>
55 #include <vm/vm_reserv.h>
56 #include <vm/vm_extern.h>
59 #include <vm/uma_int.h>
61 #include <linux/gfp.h>
63 #include <linux/preempt.h>
65 #if defined(__amd64__) || defined(__aarch64__) || defined(__riscv)
66 #define LINUXKPI_HAVE_DMAP
68 #undef LINUXKPI_HAVE_DMAP
72 linux_page_address(struct page *page)
75 if (page->object != kmem_object && page->object != kernel_object) {
76 #ifdef LINUXKPI_HAVE_DMAP
77 return ((void *)PHYS_TO_DMAP(VM_PAGE_TO_PHYS(page)));
82 return ((void *)(uintptr_t)(VM_MIN_KERNEL_ADDRESS +
83 IDX_TO_OFF(page->pindex)));
87 linux_alloc_pages(gfp_t flags, unsigned int order)
89 #ifdef LINUXKPI_HAVE_DMAP
90 unsigned long npages = 1UL << order;
91 int req = (flags & M_ZERO) ? (VM_ALLOC_ZERO | VM_ALLOC_NOOBJ |
92 VM_ALLOC_NORMAL) : (VM_ALLOC_NOOBJ | VM_ALLOC_NORMAL);
95 if (order == 0 && (flags & GFP_DMA32) == 0) {
96 page = vm_page_alloc(NULL, 0, req);
100 vm_paddr_t pmax = (flags & GFP_DMA32) ?
101 BUS_SPACE_MAXADDR_32BIT : BUS_SPACE_MAXADDR;
103 page = vm_page_alloc_contig(NULL, 0, req,
104 npages, 0, pmax, PAGE_SIZE, 0, VM_MEMATTR_DEFAULT);
107 if (flags & M_WAITOK) {
108 if (!vm_page_reclaim_contig(req,
109 npages, 0, pmax, PAGE_SIZE, 0)) {
118 if (flags & M_ZERO) {
121 for (x = 0; x != npages; x++) {
122 vm_page_t pgo = page + x;
124 if ((pgo->flags & PG_ZERO) == 0)
132 vaddr = linux_alloc_kmem(flags, order);
136 page = PHYS_TO_VM_PAGE(vtophys((void *)vaddr));
138 KASSERT(vaddr == (vm_offset_t)page_address(page),
139 ("Page address mismatch"));
145 linux_free_pages(vm_page_t page, unsigned int order)
147 #ifdef LINUXKPI_HAVE_DMAP
148 unsigned long npages = 1UL << order;
151 for (x = 0; x != npages; x++) {
152 vm_page_t pgo = page + x;
161 vaddr = (vm_offset_t)page_address(page);
163 linux_free_kmem(vaddr, order);
168 linux_alloc_kmem(gfp_t flags, unsigned int order)
170 size_t size = ((size_t)PAGE_SIZE) << order;
173 if ((flags & GFP_DMA32) == 0) {
174 addr = kmem_malloc(kmem_arena, size, flags & GFP_NATIVE_MASK);
176 addr = kmem_alloc_contig(kmem_arena, size,
177 flags & GFP_NATIVE_MASK, 0, BUS_SPACE_MAXADDR_32BIT,
178 PAGE_SIZE, 0, VM_MEMATTR_DEFAULT);
184 linux_free_kmem(vm_offset_t addr, unsigned int order)
186 size_t size = ((size_t)PAGE_SIZE) << order;
188 kmem_free(kmem_arena, addr, size);
192 linux_get_user_pages_internal(vm_map_t map, unsigned long start, int nr_pages,
193 int write, struct page **pages)
200 prot = write ? (VM_PROT_READ | VM_PROT_WRITE) : VM_PROT_READ;
201 len = ((size_t)nr_pages) << PAGE_SHIFT;
202 count = vm_fault_quick_hold_pages(map, start, len, prot, pages, nr_pages);
206 for (i = 0; i != nr_pages; i++) {
207 struct page *pg = pages[i];
217 __get_user_pages_fast(unsigned long start, int nr_pages, int write,
227 if (nr_pages == 0 || in_interrupt())
230 MPASS(pages != NULL);
232 map = &curthread->td_proc->p_vmspace->vm_map;
233 end = start + (((size_t)nr_pages) << PAGE_SHIFT);
234 if (start < vm_map_min(map) || end > vm_map_max(map))
236 prot = write ? (VM_PROT_READ | VM_PROT_WRITE) : VM_PROT_READ;
237 for (count = 0, mp = pages, va = start; va < end;
238 mp++, va += PAGE_SIZE, count++) {
239 *mp = pmap_extract_and_hold(map->pmap, va, prot);
247 if ((prot & VM_PROT_WRITE) != 0 &&
248 (*mp)->dirty != VM_PAGE_BITS_ALL) {
250 * Explicitly dirty the physical page. Otherwise, the
251 * caller's changes may go unnoticed because they are
252 * performed through an unmanaged mapping or by a DMA
255 * The object lock is not held here.
256 * See vm_page_clear_dirty_mask().
265 get_user_pages_remote(struct task_struct *task, struct mm_struct *mm,
266 unsigned long start, unsigned long nr_pages, int gup_flags,
267 struct page **pages, struct vm_area_struct **vmas)
271 map = &task->task_thread->td_proc->p_vmspace->vm_map;
272 return (linux_get_user_pages_internal(map, start, nr_pages,
273 !!(gup_flags & FOLL_WRITE), pages));
277 get_user_pages(unsigned long start, unsigned long nr_pages, int gup_flags,
278 struct page **pages, struct vm_area_struct **vmas)
282 map = &curthread->td_proc->p_vmspace->vm_map;
283 return (linux_get_user_pages_internal(map, start, nr_pages,
284 !!(gup_flags & FOLL_WRITE), pages));
288 is_vmalloc_addr(const void *addr)
290 return (vtoslab((vm_offset_t)addr & ~UMA_SLAB_MASK) != NULL);