4.4BSD/usr/src/sys/sparc/include/pmap.h
/*
* Copyright (c) 1992, 1993
* The Regents of the University of California. All rights reserved.
*
* This software was developed by the Computer Systems Engineering group
* at Lawrence Berkeley Laboratory under DARPA contract BG 91-66 and
* contributed to Berkeley.
*
* All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Lawrence Berkeley Laboratory.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)pmap.h 8.1 (Berkeley) 6/11/93
*
* from: $Header: pmap.h,v 1.11 93/05/25 10:36:09 torek Exp $
*/
#ifndef _SPARC_PMAP_H_
#define _SPARC_PMAP_H_
#include <machine/pte.h>
/*
* Pmap structure.
*
* The pmap structure really comes in two variants, one---a single
* instance---for kernel virtual memory and the other---up to nproc
* instances---for user virtual memory. Unfortunately, we have to mash
* both into the same structure. Fortunately, they are almost the same.
*
* The kernel begins at 0xf8000000 and runs to 0xffffffff (although
* some of this is not actually used). Kernel space, including DVMA
* space (for now?), is mapped identically into all user contexts.
* There is no point in duplicating this mapping in each user process
* so they do not appear in the user structures.
*
* User space begins at 0x00000000 and runs through 0x1fffffff,
* then has a `hole', then resumes at 0xe0000000 and runs until it
* hits the kernel space at 0xf8000000. This can be mapped
* contiguously by ignorning the top two bits and pretending the
* space goes from 0 to 37ffffff. Typically the lower range is
* used for text+data and the upper for stack, but the code here
* makes no such distinction.
*
* Since each virtual segment covers 256 kbytes, the user space
* requires 3584 segments, while the kernel (including DVMA) requires
* only 512 segments.
*
* The segment map entry for virtual segment vseg is offset in
* pmap->pm_rsegmap by 0 if pmap is not the kernel pmap, or by
* NUSEG if it is. We keep a pointer called pmap->pm_segmap
* pre-offset by this value. pmap->pm_segmap thus contains the
* values to be loaded into the user portion of the hardware segment
* map so as to reach the proper PMEGs within the MMU. The kernel
* mappings are `set early' and are always valid in every context
* (every change is always propagated immediately).
*
* The PMEGs within the MMU are loaded `on demand'; when a PMEG is
* taken away from context `c', the pmap for context c has its
* corresponding pm_segmap[vseg] entry marked invalid (the MMU segment
* map entry is also made invalid at the same time). Thus
* pm_segmap[vseg] is the `invalid pmeg' number (127 or 511) whenever
* the corresponding PTEs are not actually in the MMU. On the other
* hand, pm_pte[vseg] is NULL only if no pages in that virtual segment
* are in core; otherwise it points to a copy of the 32 or 64 PTEs that
* must be loaded in the MMU in order to reach those pages.
* pm_npte[vseg] counts the number of valid pages in each vseg.
*
* XXX performance: faster to count valid bits?
*
* The kernel pmap cannot malloc() PTEs since malloc() will sometimes
* allocate a new virtual segment. Since kernel mappings are never
* `stolen' out of the the MMU, we just keep all its PTEs there, and
* have no software copies. Its mmu entries are nonetheless kept on lists
* so that the code that fiddles with mmu lists has something to fiddle.
*/
#define NKSEG ((int)((-(unsigned)KERNBASE) / NBPSG)) /* i.e., 512 */
#define NUSEG (4096 - NKSEG) /* i.e., 3584 */
/* data appearing in both user and kernel pmaps */
struct pmap_common {
union ctxinfo *pmc_ctx; /* current context, if any */
int pmc_ctxnum; /* current context's number */
#if NCPUS > 1
simple_lock_data_t pmc_lock; /* spinlock */
#endif
int pmc_refcount; /* just what it says */
struct mmuentry *pmc_mmuforw; /* pmap pmeg chain */
struct mmuentry **pmc_mmuback; /* (two way street) */
pmeg_t *pmc_segmap; /* points to pm_rsegmap per above */
u_char *pmc_npte; /* points to pm_rnpte */
int **pmc_pte; /* points to pm_rpte */
};
/* data appearing only in user pmaps */
struct pmap {
struct pmap_common pmc;
pmeg_t pm_rsegmap[NUSEG]; /* segment map */
u_char pm_rnpte[NUSEG]; /* number of valid PTEs per seg */
int *pm_rpte[NUSEG]; /* points to PTEs for valid segments */
};
/* data appearing only in the kernel pmap */
struct kpmap {
struct pmap_common pmc;
pmeg_t pm_rsegmap[NKSEG]; /* segment map */
u_char pm_rnpte[NKSEG]; /* number of valid PTEs per kseg */
int *pm_rpte[NKSEG]; /* always NULL */
};
#define pm_ctx pmc.pmc_ctx
#define pm_ctxnum pmc.pmc_ctxnum
#define pm_lock pmc.pmc_lock
#define pm_refcount pmc.pmc_refcount
#define pm_mmuforw pmc.pmc_mmuforw
#define pm_mmuback pmc.pmc_mmuback
#define pm_segmap pmc.pmc_segmap
#define pm_npte pmc.pmc_npte
#define pm_pte pmc.pmc_pte
#ifdef KERNEL
typedef struct pmap *pmap_t;
#define PMAP_NULL ((pmap_t)0)
extern struct kpmap kernel_pmap_store;
#define kernel_pmap ((struct pmap *)(&kernel_pmap_store))
#define PMAP_ACTIVATE(pmap, pcb, iscurproc)
#define PMAP_DEACTIVATE(pmap, pcb)
/*
* Since PTEs also contain type bits, we have to have some way
* to tell pmap_enter `this is an IO page' or `this is not to
* be cached'. Since physical addresses are always aligned, we
* can do this with the low order bits.
*
* The ordering below is important: PMAP_PGTYPE << PG_TNC must give
* exactly the PG_NC and PG_TYPE bits.
*/
#define PMAP_OBIO 1 /* tells pmap_enter to use PG_OBIO */
#define PMAP_VME16 2 /* etc */
#define PMAP_VME32 3 /* etc */
#define PMAP_NC 4 /* tells pmap_enter to set PG_NC */
#define PMAP_TNC 7 /* mask to get PG_TYPE & PG_NC */
#endif /* KERNEL */
#endif /* _SPARC_PMAP_H_ */