xref: /minix/sys/arch/i386/include/pmap.h (revision 0a6a1f1d)
1 /*	$NetBSD: pmap.h,v 1.117 2014/04/21 19:12:11 christos Exp $	*/
2 
3 /*
4  * Copyright (c) 1997 Charles D. Cranor and Washington University.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
17  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
20  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
21  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
25  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26  */
27 
28 /*
29  * Copyright (c) 2001 Wasabi Systems, Inc.
30  * All rights reserved.
31  *
32  * Written by Frank van der Linden for Wasabi Systems, Inc.
33  *
34  * Redistribution and use in source and binary forms, with or without
35  * modification, are permitted provided that the following conditions
36  * are met:
37  * 1. Redistributions of source code must retain the above copyright
38  *    notice, this list of conditions and the following disclaimer.
39  * 2. Redistributions in binary form must reproduce the above copyright
40  *    notice, this list of conditions and the following disclaimer in the
41  *    documentation and/or other materials provided with the distribution.
42  * 3. All advertising materials mentioning features or use of this software
43  *    must display the following acknowledgement:
44  *      This product includes software developed for the NetBSD Project by
45  *      Wasabi Systems, Inc.
46  * 4. The name of Wasabi Systems, Inc. may not be used to endorse
47  *    or promote products derived from this software without specific prior
48  *    written permission.
49  *
50  * THIS SOFTWARE IS PROVIDED BY WASABI SYSTEMS, INC. ``AS IS'' AND
51  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
52  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
53  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL WASABI SYSTEMS, INC
54  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
55  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
56  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
57  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
58  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
59  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
60  * POSSIBILITY OF SUCH DAMAGE.
61  */
62 
63 #ifndef	_I386_PMAP_H_
64 #define	_I386_PMAP_H_
65 
66 #if defined(_KERNEL_OPT)
67 #include "opt_user_ldt.h"
68 #include "opt_xen.h"
69 #endif
70 
71 #include <sys/atomic.h>
72 
73 #include <i386/pte.h>
74 #include <machine/segments.h>
75 #if defined(_KERNEL)
76 #include <machine/cpufunc.h>
77 #endif
78 
79 #include <uvm/uvm_object.h>
80 #ifdef XEN
81 #include <xen/xenfunc.h>
82 #include <xen/xenpmap.h>
83 #endif /* XEN */
84 
85 /*
86  * see pte.h for a description of i386 MMU terminology and hardware
87  * interface.
88  *
89  * a pmap describes a processes' 4GB virtual address space.  when PAE
90  * is not in use, this virtual address space can be broken up into 1024 4MB
91  * regions which are described by PDEs in the PDP.  the PDEs are defined as
92  * follows:
93  *
94  * (ranges are inclusive -> exclusive, just like vm_map_entry start/end)
95  * (the following assumes that KERNBASE is 0xc0000000)
96  *
97  * PDE#s	VA range		usage
98  * 0->766	0x0 -> 0xbfc00000	user address space
99  * 767		0xbfc00000->		recursive mapping of PDP (used for
100  *			0xc0000000	linear mapping of PTPs)
101  * 768->1023	0xc0000000->		kernel address space (constant
102  *			0xffc00000	across all pmap's/processes)
103  *			<end>
104  *
105  *
106  * note: a recursive PDP mapping provides a way to map all the PTEs for
107  * a 4GB address space into a linear chunk of virtual memory.  in other
108  * words, the PTE for page 0 is the first int mapped into the 4MB recursive
109  * area.  the PTE for page 1 is the second int.  the very last int in the
110  * 4MB range is the PTE that maps VA 0xfffff000 (the last page in a 4GB
111  * address).
112  *
113  * all pmap's PD's must have the same values in slots 768->1023 so that
114  * the kernel is always mapped in every process.  these values are loaded
115  * into the PD at pmap creation time.
116  *
117  * at any one time only one pmap can be active on a processor.  this is
118  * the pmap whose PDP is pointed to by processor register %cr3.  this pmap
119  * will have all its PTEs mapped into memory at the recursive mapping
120  * point (slot #767 as show above).  when the pmap code wants to find the
121  * PTE for a virtual address, all it has to do is the following:
122  *
123  * address of PTE = (767 * 4MB) + (VA / PAGE_SIZE) * sizeof(pt_entry_t)
124  *                = 0xbfc00000 + (VA / 4096) * 4
125  *
126  * what happens if the pmap layer is asked to perform an operation
127  * on a pmap that is not the one which is currently active?  in that
128  * case we temporarily load this pmap, perform the operation, and mark
129  * the currently active one as pending lazy reload.
130  *
131  * the following figure shows the effects of the recursive PDP mapping:
132  *
133  *   PDP (%cr3)
134  *   +----+
135  *   |   0| -> PTP#0 that maps VA 0x0 -> 0x400000
136  *   |    |
137  *   |    |
138  *   | 767| -> points back to PDP (%cr3) mapping VA 0xbfc00000 -> 0xc0000000
139  *   | 768| -> first kernel PTP (maps 0xc0000000 -> 0xc0400000)
140  *   |    |
141  *   +----+
142  *
143  * note that the PDE#767 VA (0xbfc00000) is defined as "PTE_BASE"
144  *
145  * starting at VA 0xbfc00000 the current active PDP (%cr3) acts as a
146  * PTP:
147  *
148  * PTP#767 == PDP(%cr3) => maps VA 0xbfc00000 -> 0xc0000000
149  *   +----+
150  *   |   0| -> maps the contents of PTP#0 at VA 0xbfc00000->0xbfc01000
151  *   |    |
152  *   |    |
153  *   | 767| -> maps contents of PTP#767 (the PDP) at VA 0xbfeff000
154  *   | 768| -> maps contents of first kernel PTP
155  *   |    |
156  *   |1023|
157  *   +----+
158  *
159  * note that mapping of the PDP at PTP#767's VA (0xbfeff000) is
160  * defined as "PDP_BASE".... within that mapping there are two
161  * defines:
162  *   "PDP_PDE" (0xbfeffbfc) is the VA of the PDE in the PDP
163  *      which points back to itself.
164  *
165  * - PAE support -
166  * ---------------
167  *
168  * PAE adds another layer of indirection during address translation, breaking
169  * up the translation process in 3 different levels:
170  * - L3 page directory, containing 4 * 64-bits addresses (index determined by
171  * bits [31:30] from the virtual address). This breaks up the address space
172  * in 4 1GB regions.
173  * - the PD (L2), containing 512 64-bits addresses, breaking each L3 region
174  * in 512 * 2MB regions.
175  * - the PT (L1), also containing 512 64-bits addresses (at L1, the size of
176  * the pages is still 4K).
177  *
178  * The kernel virtual space is mapped by the last entry in the L3 page,
179  * the first 3 entries mapping the user VA space.
180  *
181  * Because the L3 has only 4 entries of 1GB each, we can't use recursive
182  * mappings at this level for PDP_PDE (this would eat up 2 of the 4GB
183  * virtual space). There are also restrictions imposed by Xen on the
184  * last entry of the L3 PD (reference count to this page cannot be
185  * bigger than 1), which makes it hard to use one L3 page per pmap to
186  * switch between pmaps using %cr3.
187  *
188  * As such, each CPU gets its own L3 page that is always loaded into its %cr3
189  * (ci_pae_l3_pd in the associated cpu_info struct). We claim that the VM has
190  * only a 2-level PTP (similar to the non-PAE case). L2 PD is now 4 contiguous
191  * pages long (corresponding to the 4 entries of the L3), and the different
192  * index/slots (like PDP_PDE) are adapted accordingly.
193  *
194  * Kernel space remains in L3[3], L3[0-2] maps the user VA space. Switching
195  * between pmaps consists in modifying the first 3 entries of the CPU's L3 page.
196  *
197  * PTE_BASE will need 4 entries in the L2 PD pages to map the L2 pages
198  * recursively.
199  *
200  * In addition, for Xen, we can't recursively map L3[3] (Xen wants the ref
201  * count on this page to be exactly one), so we use a shadow PD page for
202  * the last L2 PD. The shadow page could be static too, but to make pm_pdir[]
203  * contiguous we'll allocate/copy one page per pmap.
204  */
205 
206 /*
207  * Mask to get rid of the sign-extended part of addresses.
208  */
209 #define VA_SIGN_MASK		0
210 #define VA_SIGN_NEG(va)		((va) | VA_SIGN_MASK)
211 /*
212  * XXXfvdl this one's not right.
213  */
214 #define VA_SIGN_POS(va)		((va) & ~VA_SIGN_MASK)
215 
216 /*
217  * the following defines identify the slots used as described above.
218  */
219 #ifdef PAE
220 #define L2_SLOT_PTE	(KERNBASE/NBPD_L2-4) /* 1532: for recursive PDP map */
221 #define L2_SLOT_KERN	(KERNBASE/NBPD_L2)   /* 1536: start of kernel space */
222 #else /* PAE */
223 #define L2_SLOT_PTE	(KERNBASE/NBPD_L2-1) /* 767: for recursive PDP map */
224 #define L2_SLOT_KERN	(KERNBASE/NBPD_L2)   /* 768: start of kernel space */
225 #endif /* PAE */
226 
227 #define	L2_SLOT_KERNBASE L2_SLOT_KERN
228 
229 #define PDIR_SLOT_KERN	L2_SLOT_KERN
230 #define PDIR_SLOT_PTE	L2_SLOT_PTE
231 
232 /*
233  * the following defines give the virtual addresses of various MMU
234  * data structures:
235  * PTE_BASE: the base VA of the linear PTE mappings
236  * PDP_BASE: the base VA of the recursive mapping of the PDP
237  * PDP_PDE: the VA of the PDE that points back to the PDP
238  */
239 
240 #define PTE_BASE  ((pt_entry_t *) (PDIR_SLOT_PTE * NBPD_L2))
241 
242 #define L1_BASE		PTE_BASE
243 
244 #define L2_BASE ((pd_entry_t *)((char *)L1_BASE + L2_SLOT_PTE * NBPD_L1))
245 
246 #define PDP_PDE		(L2_BASE + PDIR_SLOT_PTE)
247 
248 #define PDP_BASE	L2_BASE
249 
250 /* largest value (-1 for APTP space) */
251 #define NKL2_MAX_ENTRIES	(NTOPLEVEL_PDES - (KERNBASE/NBPD_L2) - 1)
252 #define NKL1_MAX_ENTRIES	(unsigned long)(NKL2_MAX_ENTRIES * NPDPG)
253 
254 #define NKL2_KIMG_ENTRIES	0	/* XXX unused */
255 
256 #define NKL2_START_ENTRIES	0	/* XXX computed on runtime */
257 #define NKL1_START_ENTRIES	0	/* XXX unused */
258 
259 #ifndef XEN
260 #define NTOPLEVEL_PDES		(PAGE_SIZE * PDP_SIZE / (sizeof (pd_entry_t)))
261 #else	/* !XEN */
262 #ifdef  PAE
263 #define NTOPLEVEL_PDES		1964	/* 1964-2047 reserved by Xen */
264 #else	/* PAE */
265 #define NTOPLEVEL_PDES		1008	/* 1008-1023 reserved by Xen */
266 #endif	/* PAE */
267 #endif  /* !XEN */
268 #define NPDPG			(PAGE_SIZE / sizeof (pd_entry_t))
269 
270 #define PTP_MASK_INITIALIZER	{ L1_FRAME, L2_FRAME }
271 #define PTP_SHIFT_INITIALIZER	{ L1_SHIFT, L2_SHIFT }
272 #define NKPTP_INITIALIZER	{ NKL1_START_ENTRIES, NKL2_START_ENTRIES }
273 #define NKPTPMAX_INITIALIZER	{ NKL1_MAX_ENTRIES, NKL2_MAX_ENTRIES }
274 #define NBPD_INITIALIZER	{ NBPD_L1, NBPD_L2 }
275 #define PDES_INITIALIZER	{ L2_BASE }
276 
277 #define PTP_LEVELS	2
278 
279 /*
280  * PG_AVAIL usage: we make use of the ignored bits of the PTE
281  */
282 
283 #define PG_W		PG_AVAIL1	/* "wired" mapping */
284 #define PG_PVLIST	PG_AVAIL2	/* mapping has entry on pvlist */
285 #define PG_X		PG_AVAIL3	/* executable mapping */
286 
287 /*
288  * Number of PTE's per cache line.  4 byte pte, 32-byte cache line
289  * Used to avoid false sharing of cache lines.
290  */
291 #ifdef PAE
292 #define NPTECL		4
293 #else
294 #define NPTECL		8
295 #endif
296 
297 #include <x86/pmap.h>
298 
299 #ifndef XEN
300 #define pmap_pa2pte(a)			(a)
301 #define pmap_pte2pa(a)			((a) & PG_FRAME)
302 #define pmap_pte_set(p, n)		do { *(p) = (n); } while (0)
303 #define pmap_pte_flush()		/* nothing */
304 
305 #ifdef PAE
306 #define pmap_pte_cas(p, o, n)		atomic_cas_64((p), (o), (n))
307 #define pmap_pte_testset(p, n)		\
308     atomic_swap_64((volatile uint64_t *)p, n)
309 #define pmap_pte_setbits(p, b)		\
310     atomic_or_64((volatile uint64_t *)p, b)
311 #define pmap_pte_clearbits(p, b)	\
312     atomic_and_64((volatile uint64_t *)p, ~(b))
313 #else /* PAE */
314 #define pmap_pte_cas(p, o, n)		atomic_cas_32((p), (o), (n))
315 #define pmap_pte_testset(p, n)		\
316     atomic_swap_ulong((volatile unsigned long *)p, n)
317 #define pmap_pte_setbits(p, b)		\
318     atomic_or_ulong((volatile unsigned long *)p, b)
319 #define pmap_pte_clearbits(p, b)	\
320     atomic_and_ulong((volatile unsigned long *)p, ~(b))
321 #endif /* PAE */
322 
323 #else /* XEN */
324 extern kmutex_t pte_lock;
325 
326 static __inline pt_entry_t
pmap_pa2pte(paddr_t pa)327 pmap_pa2pte(paddr_t pa)
328 {
329 	return (pt_entry_t)xpmap_ptom_masked(pa);
330 }
331 
332 static __inline paddr_t
pmap_pte2pa(pt_entry_t pte)333 pmap_pte2pa(pt_entry_t pte)
334 {
335 	return xpmap_mtop_masked(pte & PG_FRAME);
336 }
337 static __inline void
pmap_pte_set(pt_entry_t * pte,pt_entry_t npte)338 pmap_pte_set(pt_entry_t *pte, pt_entry_t npte)
339 {
340 	int s = splvm();
341 	xpq_queue_pte_update(xpmap_ptetomach(pte), npte);
342 	splx(s);
343 }
344 
345 static __inline pt_entry_t
pmap_pte_cas(volatile pt_entry_t * ptep,pt_entry_t o,pt_entry_t n)346 pmap_pte_cas(volatile pt_entry_t *ptep, pt_entry_t o, pt_entry_t n)
347 {
348 	pt_entry_t opte;
349 
350 	mutex_enter(&pte_lock);
351 	opte = *ptep;
352 	if (opte == o) {
353 		xpq_queue_pte_update(xpmap_ptetomach(__UNVOLATILE(ptep)), n);
354 		xpq_flush_queue();
355 	}
356 	mutex_exit(&pte_lock);
357 	return opte;
358 }
359 
360 static __inline pt_entry_t
pmap_pte_testset(volatile pt_entry_t * pte,pt_entry_t npte)361 pmap_pte_testset(volatile pt_entry_t *pte, pt_entry_t npte)
362 {
363 	pt_entry_t opte;
364 
365 	mutex_enter(&pte_lock);
366 	opte = *pte;
367 	xpq_queue_pte_update(xpmap_ptetomach(__UNVOLATILE(pte)),
368 	    npte);
369 	xpq_flush_queue();
370 	mutex_exit(&pte_lock);
371 	return opte;
372 }
373 
374 static __inline void
pmap_pte_setbits(volatile pt_entry_t * pte,pt_entry_t bits)375 pmap_pte_setbits(volatile pt_entry_t *pte, pt_entry_t bits)
376 {
377 	mutex_enter(&pte_lock);
378 	xpq_queue_pte_update(xpmap_ptetomach(__UNVOLATILE(pte)), (*pte) | bits);
379 	xpq_flush_queue();
380 	mutex_exit(&pte_lock);
381 }
382 
383 static __inline void
pmap_pte_clearbits(volatile pt_entry_t * pte,pt_entry_t bits)384 pmap_pte_clearbits(volatile pt_entry_t *pte, pt_entry_t bits)
385 {
386 	mutex_enter(&pte_lock);
387 	xpq_queue_pte_update(xpmap_ptetomach(__UNVOLATILE(pte)),
388 	    (*pte) & ~bits);
389 	xpq_flush_queue();
390 	mutex_exit(&pte_lock);
391 }
392 
393 static __inline void
pmap_pte_flush(void)394 pmap_pte_flush(void)
395 {
396 	int s = splvm();
397 	xpq_flush_queue();
398 	splx(s);
399 }
400 
401 #endif
402 
403 struct vm_map;
404 struct trapframe;
405 struct pcb;
406 
407 int	pmap_exec_fixup(struct vm_map *, struct trapframe *, struct pcb *);
408 void	pmap_ldt_cleanup(struct lwp *);
409 
410 #include <x86/pmap_pv.h>
411 
412 #define	__HAVE_VM_PAGE_MD
413 #define	VM_MDPAGE_INIT(pg) \
414 	memset(&(pg)->mdpage, 0, sizeof((pg)->mdpage)); \
415 	PMAP_PAGE_INIT(&(pg)->mdpage.mp_pp)
416 
417 struct vm_page_md {
418 	struct pmap_page mp_pp;
419 };
420 
421 #endif	/* _I386_PMAP_H_ */
422