summaryrefslogtreecommitdiff
path: root/sys
diff options
context:
space:
mode:
authorleo <leo@NetBSD.org>1998-05-07 07:25:51 +0000
committerleo <leo@NetBSD.org>1998-05-07 07:25:51 +0000
commit4eb4a78df8a4899f03ff271eccb41c4bd217d7c4 (patch)
tree7ac223cab0a49721ea9b03bfe1d4729c956c3795 /sys
parent10a773225e6702770ab46fcefb65474f8687fced (diff)
Implement MACHINE_NEW_NONCONTIG. Implementation is based on Ignatios' amiga
implementation and Chuck's conversion description.
Diffstat (limited to 'sys')
-rw-r--r--sys/arch/atari/atari/machdep.c23
-rw-r--r--sys/arch/atari/atari/pmap.c391
-rw-r--r--sys/arch/atari/conf/std.atari4
-rw-r--r--sys/arch/atari/include/pmap.h9
-rw-r--r--sys/arch/atari/include/vmparam.h18
5 files changed, 208 insertions, 237 deletions
diff --git a/sys/arch/atari/atari/machdep.c b/sys/arch/atari/atari/machdep.c
index adb9085e09a..04f869b8845 100644
--- a/sys/arch/atari/atari/machdep.c
+++ b/sys/arch/atari/atari/machdep.c
@@ -1,4 +1,4 @@
-/* $NetBSD: machdep.c,v 1.60 1998/03/13 21:05:10 leo Exp $ */
+/* $NetBSD: machdep.c,v 1.61 1998/05/07 07:25:51 leo Exp $ */
/*
* Copyright (c) 1988 University of Utah.
@@ -97,8 +97,6 @@ static void netintr __P((void));
void straymfpint __P((int, u_short));
void straytrap __P((int, u_short));
-extern vm_offset_t avail_end;
-
/*
* Declare these as initialized data so we can patch them.
*/
@@ -107,7 +105,8 @@ int bufpages = BUFPAGES;
#else
int bufpages = 0;
#endif
-caddr_t msgbufaddr;
+caddr_t msgbufaddr;
+vm_offset_t msgbufpa;
int physmem = MAXMEM; /* max supported memory, changes to actual */
/*
@@ -164,7 +163,7 @@ cpu_startup()
#endif
vm_offset_t minaddr, maxaddr;
vm_size_t size = 0;
- u_long memsize;
+ extern vm_size_t mem_size; /* from pmap.c */
/*
* Initialize error message buffer (at end of core).
@@ -172,10 +171,13 @@ cpu_startup()
#ifdef DEBUG
pmapdebug = 0;
#endif
- /* avail_end was pre-decremented in pmap_bootstrap to compensate */
+ /*
+ * pmap_bootstrap has positioned this at the end of kernel
+ * memory segment - map and initialize it now.
+ */
for (i = 0; i < btoc(MSGBUFSIZE); i++)
pmap_enter(pmap_kernel(), (vm_offset_t)msgbufaddr + i * NBPG,
- avail_end + i * NBPG, VM_PROT_ALL, TRUE);
+ msgbufpa + i * NBPG, VM_PROT_ALL, TRUE);
initmsgbuf(msgbufaddr, m68k_round_page(MSGBUFSIZE));
/*
@@ -184,12 +186,7 @@ cpu_startup()
printf(version);
identifycpu();
- for (i = memsize = 0; i < NMEM_SEGS; i++) {
- if (boot_segs[i].start == boot_segs[i].end)
- break;
- memsize += boot_segs[i].end - boot_segs[i].start;
- }
- printf("real mem = %ld (%ld pages)\n", memsize, memsize/NBPG);
+ printf("real mem = %ld (%ld pages)\n", mem_size, mem_size/NBPG);
/*
* Allocate space for system data structures.
diff --git a/sys/arch/atari/atari/pmap.c b/sys/arch/atari/atari/pmap.c
index 213376d944b..b70687b6318 100644
--- a/sys/arch/atari/atari/pmap.c
+++ b/sys/arch/atari/atari/pmap.c
@@ -1,4 +1,4 @@
-/* $NetBSD: pmap.c,v 1.24 1998/01/06 07:03:01 thorpej Exp $ */
+/* $NetBSD: pmap.c,v 1.25 1998/05/07 07:25:52 leo Exp $ */
/*
* Copyright (c) 1991 Regents of the University of California.
@@ -200,7 +200,17 @@ int protection_codes[8];
/*
* Kernel page table page management.
+ *
+ * One additional page of KPT allows for 16 MB of virtual buffer cache.
+ * A GENERIC kernel allocates this for 2 MB of real buffer cache,
+ * which in turn is allocated for 38 MB of RAM.
+ * We add one per 16 MB of RAM to allow for tuning the machine-independent
+ * options.
*/
+#ifndef NKPTADDSHIFT
+#define NKPTADDSHIFT 24
+#endif
+
struct kpt_page {
struct kpt_page *kpt_next; /* link on either used or free list */
vm_offset_t kpt_va; /* always valid kernel VA */
@@ -226,13 +236,11 @@ vm_size_t Sysptsize = VM_KERNEL_PT_PAGES + 4 / NPTEPG;
struct pmap kernel_pmap_store;
vm_map_t pt_map;
-vm_offset_t avail_start; /* PA of first available physical page */
-vm_offset_t avail_end; /* PA of last available physical page */
vm_size_t mem_size; /* memory size in bytes */
+vm_offset_t avail_end; /* PA of last available physical page */
vm_offset_t virtual_avail; /* VA of first avail page (after kernel bss)*/
vm_offset_t virtual_end; /* VA of last avail page (end of kernel AS) */
-vm_offset_t vm_first_phys; /* PA of first managed page */
-vm_offset_t vm_last_phys; /* PA just past last managed page */
+int page_cnt; /* number of pages managed by the VM system */
boolean_t pmap_initialized = FALSE; /* Has pmap_init completed? */
char *pmap_attributes; /* reference and modify bits */
TAILQ_HEAD(pv_page_list, pv_page) pv_page_freelist;
@@ -242,26 +250,15 @@ static int pmap_ishift; /* segment table index shift */
int protostfree; /* prototype (default) free ST map */
#endif
-extern caddr_t msgbufaddr;
-
-#ifdef MACHINE_NONCONTIG
-static vm_offset_t avail_next;
-static vm_size_t avail_remaining;
-#endif
+extern caddr_t msgbufaddr;
+extern vm_offset_t msgbufpa;
static boolean_t pmap_testbit __P((vm_offset_t, int));
static void pmap_enter_ptpage __P((pmap_t, vm_offset_t));
static struct pv_entry* pmap_alloc_pv __P((void));
static void pmap_free_pv __P((struct pv_entry *));
static void atari_protection_init __P((void));
-static int pmap_isvalidphys __P((vm_offset_t));
-
-#ifdef MACHINE_NONCONTIG
-#define pmap_valid_page(pa) (pmap_initialized && pmap_isvalidphys(pa))
-#else
-#define pmap_valid_page(pa) (pmap_initialized && pa >= vm_first_phys && \
- pa < vm_last_phys)
-#endif
+static void pmap_collect1 __P((pmap_t, vm_offset_t, vm_offset_t));
/*
* All those kernel PT submaps that BSD is so fond of
@@ -269,6 +266,25 @@ static int pmap_isvalidphys __P((vm_offset_t));
caddr_t CADDR1, CADDR2;
u_int *CMAP1, *CMAP2, *vmpte, *msgbufmap;
+#define PAGE_IS_MANAGED(pa) (pmap_initialized \
+ && vm_physseg_find(atop((pa)), NULL) != -1)
+
+#define pa_to_pvh(pa) \
+({ \
+ int bank_, pg_; \
+ \
+ bank_ = vm_physseg_find(atop((pa)), &pg_); \
+ &vm_physmem[bank_].pmseg.pvent[pg_]; \
+})
+
+#define pa_to_attribute(pa) \
+({ \
+ int bank_, pg_; \
+ \
+ bank_ = vm_physseg_find(atop((pa)), &pg_); \
+ &vm_physmem[bank_].pmseg.attrs[pg_]; \
+})
+
/*
* The preallocated virtual memory range used by the I/O area. Their
* values are passed to pmap_bootstrap().
@@ -309,20 +325,26 @@ u_int hw_addr, hw_pages;
;
/* XXX: allow for msgbuf */
usable_segs[i].end -= m68k_round_page(MSGBUFSIZE);
+ avail_end = msgbufpa = usable_segs[i].end;
- avail_start = usable_segs[0].start;
- avail_end = usable_segs[i].end;
+ /*
+ * Count physical memory
+ */
+ for (i = mem_size = 0; i < NMEM_SEGS; i++) {
+ if (boot_segs[i].start == boot_segs[i].end)
+ break;
+ mem_size += boot_segs[i].end - boot_segs[i].start;
+ }
-#ifdef MACHINE_NONCONTIG
/*
- * Setup var's for pmap_next_page()
+ * Announce available memory to the VM-system
*/
- for (i = 0, avail_remaining = 0; usable_segs[i].start; i++)
- avail_remaining += usable_segs[i].end - usable_segs[i].start;
+ for (i = 0; usable_segs[i].start; i++)
+ vm_page_physload(atop(usable_segs[i].start),
+ atop(usable_segs[i].end),
+ atop(usable_segs[i].start),
+ atop(usable_segs[i].end));
- avail_remaining >>= PGSHIFT;
- avail_next = usable_segs[0].start;
-#endif
virtual_avail = VM_MIN_KERNEL_ADDRESS + kernel_size;
virtual_end = VM_MAX_KERNEL_ADDRESS;
@@ -364,64 +386,23 @@ u_int hw_addr, hw_pages;
SYSMAP(caddr_t ,vmpte ,vmmap ,1 )
SYSMAP(caddr_t ,msgbufmap ,msgbufaddr ,btoc(MSGBUFSIZE) )
- virtual_avail = reserve_dumppages(va);
-}
+ DCIS();
-/*
- * Bootstrap memory allocator. This function allows for early dynamic
- * memory allocation until the virtual memory system has been bootstrapped.
- * After that point, either kmem_alloc or malloc should be used. This
- * function works by stealing pages from the (to be) managed page pool,
- * stealing virtual address space, then mapping the pages and zeroing them.
- *
- * It should be used from pmap_bootstrap till vm_page_startup, afterwards
- * it cannot be used, and will generate a panic if tried. Note that this
- * memory will never be freed, and in essence it is wired down.
- */
-void *
-pmap_bootstrap_alloc(size)
- int size;
-{
- extern boolean_t vm_page_startup_initialized;
- vm_offset_t val;
-
- if (vm_page_startup_initialized)
- panic("pmap_bootstrap_alloc: called after startup initialized");
- size = round_page(size);
- val = virtual_avail;
-
- virtual_avail = pmap_map(virtual_avail, avail_start,
- avail_start + size, VM_PROT_READ|VM_PROT_WRITE);
- avail_start += size;
-
- bzero((caddr_t) val, size);
- return((void *) val);
+ virtual_avail = reserve_dumppages(va);
}
-/*
- * Initialize the pmap module.
- * Called by vm_init, to initialize any structures that the pmap
- * system needs to map virtual memory.
- */
void
-#ifdef MACHINE_NONCONTIG
pmap_init()
-#else
-pmap_init(phys_start, phys_end)
- vm_offset_t phys_start, phys_end;
-#endif
{
vm_offset_t addr, addr2;
vm_size_t npg, s;
- int rv;
+ struct pv_entry *pv;
+ char *attr;
+ int rv, bank;
#ifdef DEBUG
if (pmapdebug & PDB_FOLLOW)
-#ifdef MACHINE_NONCONTIG
- printf("pmap_init(%lx, %lx)\n", avail_start, avail_end);
-#else
- printf("pmap_init(%lx, %lx)\n", phys_start, phys_end);
-#endif
+ printf("pmap_init()\n");
#endif
/*
* Now that kernel map has been allocated, we can mark as
@@ -447,8 +428,7 @@ pmap_init(phys_start, phys_end)
if (pmapdebug & PDB_INIT) {
printf("pmap_init: Sysseg %p, Sysmap %p, Sysptmap %p\n",
Sysseg, Sysmap, Sysptmap);
- printf(" pstart %lx, pend %lx, vstart %lx, vend %lx\n",
- avail_start, avail_end, virtual_avail, virtual_end);
+ printf(" vstart %lx, vend %lx\n", virtual_avail, virtual_end);
}
#endif
@@ -456,56 +436,65 @@ pmap_init(phys_start, phys_end)
* Allocate memory for random pmap data structures. Includes the
* initial segment table, pv_head_table and pmap_attributes.
*/
-#ifdef MACHINE_NONCONTIG
- {
- int i;
- for (npg = 0, i = 0; usable_segs[i].start; ++i)
- npg += atop(usable_segs[i].end - usable_segs[i].start);
- }
+ for (page_cnt = 0, bank = 0; bank < vm_nphysseg; bank++) {
+ page_cnt += vm_physmem[bank].end - vm_physmem[bank].start;
#ifdef DEBUG
- printf ("pmap_init: avail_start %08lx usable_segs[0].start %08lx"
- " npg %ld\n", avail_start, usable_segs[0].start, npg);
-#endif
-#else
- npg = atop(phys_end - phys_start);
+ printf("pmap_init: %2d: %08lx - %08lx (%10d)\n", bank,
+ vm_physmem[bank].start << PGSHIFT,
+ vm_physmem[bank].end << PGSHIFT, page_cnt << PGSHIFT);
#endif
- s = (vm_size_t)ATARI_STSIZE +
- sizeof(struct pv_entry) * npg + npg;
-
+ }
+ s = ATARI_STSIZE; /* Segtabzero */
+ s += page_cnt * sizeof(struct pv_entry); /* pv table */
+ s += page_cnt * sizeof(char); /* attribute table */
s = round_page(s);
+
addr = (vm_offset_t) kmem_alloc(kernel_map, s);
Segtabzero = (u_int *) addr;
Segtabzeropa = (u_int *) pmap_extract(pmap_kernel(), addr);
-#ifdef M68060
- if (cputype == CPU_68060) {
- addr2 = addr;
- while (addr2 < addr + ATARI_STSIZE) {
- pmap_changebit(addr2, PG_CCB, 0);
- pmap_changebit(addr2, PG_CI, 1);
- addr2 += NBPG;
- }
- DCIS();
- }
-#endif
-
addr += ATARI_STSIZE;
pv_table = (pv_entry_t) addr;
- addr += sizeof(struct pv_entry) * npg;
+ addr += page_cnt * sizeof(struct pv_entry);
+
pmap_attributes = (char *) addr;
#ifdef DEBUG
if (pmapdebug & PDB_INIT)
- printf("pmap_init: %lx bytes (%lx pgs): seg %p tbl %p attr %p\n",
- s, npg, Segtabzero, pv_table, pmap_attributes);
+ printf("pmap_init: %lx bytes: page_cnt %x s0 %p(%p) "
+ "tbl %p atr %p\n",
+ s, page_cnt, Segtabzero, Segtabzeropa,
+ pv_table, pmap_attributes);
#endif
/*
+ * Now that the pv and attribute tables have been allocated,
+ * assign them to the memory segments.
+ */
+ pv = pv_table;
+ attr = pmap_attributes;
+ for (bank = 0; bank < vm_nphysseg; bank++) {
+ npg = vm_physmem[bank].end - vm_physmem[bank].start;
+ vm_physmem[bank].pmseg.pvent = pv;
+ vm_physmem[bank].pmseg.attrs = attr;
+ pv += npg;
+ attr += npg;
+ }
+
+ /*
* Allocate physical memory for kernel PT pages and their management.
* we need enough pages to map the page tables for each process
* plus some slop.
*/
npg = howmany(((maxproc + 16) * ATARI_UPTSIZE / NPTEPG), NBPG);
- npg = min(atop(ATARI_MAX_KPTSIZE), npg);
+#ifdef NKPTADD
+ npg += NKPTADD;
+#else
+ npg += mem_size >> NKPTADDSHIFT;
+#endif
+#ifdef DEBUG
+ printf("Maxproc %d, mem_size %ld MB: allocating %ld KPT pages\n",
+ maxproc, mem_size>>20, npg);
+#endif
s = ptoa(npg) + round_page(npg * sizeof(struct kpt_page));
/*
@@ -534,13 +523,6 @@ pmap_init(phys_start, phys_end)
kpt_pages->kpt_va = addr2;
kpt_pages->kpt_pa = pmap_extract(pmap_kernel(), addr2);
-#ifdef M68060
- if (cputype == CPU_68060) {
- pmap_changebit(kpt_pages->kpt_pa, PG_CCB, 0);
- pmap_changebit(kpt_pages->kpt_pa, PG_CI, 1);
- DCIS();
- }
-#endif
} while (addr != addr2);
#ifdef DEBUG
kpt_stats.kpttotal = atop(s);
@@ -589,79 +571,35 @@ pmap_init(phys_start, phys_end)
/*
* Now it is safe to enable pv_table recording.
*/
-#ifdef MACHINE_NONCONTIG
- vm_first_phys = avail_start;
- vm_last_phys = avail_end;
-#else
- vm_first_phys = phys_start;
- vm_last_phys = phys_end;
-#endif
pmap_initialized = TRUE;
-}
-
-#ifdef MACHINE_NONCONTIG
-u_int
-pmap_free_pages()
-{
- return avail_remaining;
-}
-
-int
-pmap_next_page(addrp)
- vm_offset_t *addrp;
-{
- static int cur_seg = 0;
-
-#ifdef DEBUG
- if(!cur_seg && (avail_next == usable_segs[cur_seg].start))
- printf ("pmap_next_page: next %08lx remain %ld\n",
- avail_next, avail_remaining);
-#endif
- if (usable_segs[cur_seg].start == 0)
- return FALSE;
- if (avail_next == usable_segs[cur_seg].end) {
- avail_next = usable_segs[++cur_seg].start;
-#ifdef DEBUG
- printf ("pmap_next_page: next %08lx remain %ld\n",
- avail_next, avail_remaining);
-#endif
- }
-
- if (avail_next == 0)
- return FALSE;
- *addrp = avail_next;
- avail_next += NBPG;
- avail_remaining--;
- return TRUE;
-}
-
-int
-pmap_page_index(pa)
- vm_offset_t pa;
-{
+ /*
+ * Now that this is done, mark the pages shared with the
+ * hardware page table search as non-CCB (actually, as CI).
+ *
+ * XXX Hm. Given that this is in the kernel map, can't we just
+ * use the va's?
+ */
+#ifdef M68060
+ if (cputype == CPU_68060) {
+ kptp = kpt_free_list;
+ while (kptp) {
+ pmap_changebit(kptp->kpt_pa, PG_CCB, 0);
+ pmap_changebit(kptp->kpt_pa, PG_CI, 1);
+ kptp = kptp->kpt_next;
+ }
- struct memseg *sep = &usable_segs[0];
+ addr2 = (vm_offset_t)Segtabzeropa;
+ while (addr2 < (vm_offset_t)Segtabzeropa + ATARI_STSIZE) {
+ pmap_changebit(addr2, PG_CCB, 0);
+ pmap_changebit(addr2, PG_CI, 1);
+ addr2 += NBPG;
+ }
- while (sep->start) {
- if (pa >= sep->start && pa < sep->end)
- return (m68k_btop(pa - sep->start) + sep->first_page);
- ++sep;
+ DCIS();
}
- return -1;
-}
-
-void
-pmap_virtual_space(startp, endp)
- vm_offset_t *startp;
- vm_offset_t *endp;
-{
- *startp = virtual_avail;
- *endp = virtual_end;
+#endif
}
-#else
-#define pmap_page_index(pa) (pa_index(pa))
-#endif /* MACHINE_NONCONTIG */
struct pv_entry *
pmap_alloc_pv()
@@ -981,7 +919,7 @@ pmap_remove(pmap, sva, eva)
* Remove from the PV table (raise IPL since we
* may be called at interrupt time).
*/
- if (!pmap_valid_page(pa))
+ if (!PAGE_IS_MANAGED(pa))
continue;
pv = pa_to_pvh(pa);
ste = (int *)0;
@@ -1109,7 +1047,7 @@ pmap_remove(pmap, sva, eva)
/*
* Update saved attributes for managed page
*/
- pmap_attributes[pa_index(pa)] |= bits;
+ *pa_to_attribute(pa) |= bits;
splx(s);
}
if (flushcache) {
@@ -1145,7 +1083,7 @@ pmap_page_protect(pa, prot)
(prot == VM_PROT_NONE && (pmapdebug & PDB_REMOVE)))
printf("pmap_page_protect(%lx, %x)\n", pa, prot);
#endif
- if (!pmap_valid_page(pa))
+ if (!PAGE_IS_MANAGED(pa))
return;
switch (prot) {
@@ -1377,7 +1315,7 @@ pmap_enter(pmap, va, pa, prot, wired)
* Note that we raise IPL while manipulating pv_table
* since pmap_enter can be called at interrupt time.
*/
- if (pmap_valid_page(pa)) {
+ if (PAGE_IS_MANAGED(pa)) {
register pv_entry_t pv, npv;
int s;
@@ -1664,16 +1602,8 @@ void
pmap_collect(pmap)
pmap_t pmap;
{
- register vm_offset_t pa;
- register pv_entry_t pv;
- register int *pte;
- vm_offset_t kpa;
- int s;
+ int bank, s;
-#ifdef DEBUG
- int *ste;
- int opmapdebug = 0;
-#endif
if (pmap != pmap_kernel())
return;
@@ -1683,7 +1613,40 @@ pmap_collect(pmap)
kpt_stats.collectscans++;
#endif
s = splimp();
- for (pa = vm_first_phys; pa < vm_last_phys; pa += PAGE_SIZE) {
+
+ for (bank = 0; bank < vm_nphysseg; bank++)
+ pmap_collect1(pmap, ptoa(vm_physmem[bank].start),
+ ptoa(vm_physmem[bank].end));
+
+#ifdef notyet
+ /* Go compact and garbage-collect the pv_table. */
+ pmap_collect_pv();
+#endif
+ splx(s);
+}
+
+/*
+ * Routine: pmap_collect1()
+ *
+ * Function:
+ * Helper function for pmap_collect(). Do the actual
+ * garbage-collection of range of physical addresses.
+ */
+static void
+pmap_collect1(pmap, startpa, endpa)
+ pmap_t pmap;
+ vm_offset_t startpa, endpa;
+{
+ vm_offset_t pa;
+ struct pv_entry *pv;
+ pt_entry_t *pte;
+ vm_offset_t kpa;
+#ifdef DEBUG
+ int *ste;
+ int opmapdebug = 0;
+#endif
+
+ for (pa = startpa; pa < endpa; pa += NBPG) {
register struct kpt_page *kpt, **pkpt;
/*
@@ -1710,9 +1673,9 @@ pmap_collect(pmap)
ok:
#endif
pte = (int *)(pv->pv_va + NBPG);
- while (--pte >= (int *)pv->pv_va && *pte == PG_NV)
+ while (--pte >= (pt_entry_t *)pv->pv_va && *pte == PG_NV)
;
- if (pte >= (int *)pv->pv_va)
+ if (pte >= (pt_entry_t *)pv->pv_va)
continue;
#ifdef DEBUG
@@ -1767,7 +1730,6 @@ ok:
ste, *ste);
#endif
}
- splx(s);
}
/*
@@ -1879,7 +1841,7 @@ pmap_pageable(pmap, sva, eva, pageable)
if (!pmap_ste_v(pmap, sva))
return;
pa = pmap_pte_pa(pmap_pte(pmap, sva));
- if (!pmap_valid_page(pa))
+ if (!PAGE_IS_MANAGED(pa))
return;
pv = pa_to_pvh(pa);
if (pv->pv_ptste == NULL)
@@ -2024,7 +1986,7 @@ pmap_testbit(pa, bit)
register int *pte;
int s;
- if (!pmap_valid_page(pa))
+ if (!PAGE_IS_MANAGED(pa))
return(FALSE);
pv = pa_to_pvh(pa);
@@ -2032,7 +1994,7 @@ pmap_testbit(pa, bit)
/*
* Check saved info first
*/
- if (pmap_attributes[pa_index(pa)] & bit) {
+ if (*pa_to_attribute(pa) & bit) {
splx(s);
return(TRUE);
}
@@ -2072,7 +2034,7 @@ pmap_changebit(pa, bit, setem)
printf("pmap_changebit(%lx, %x, %s)\n",
pa, bit, setem ? "set" : "clear");
#endif
- if (!pmap_valid_page(pa))
+ if (!PAGE_IS_MANAGED(pa))
return;
pv = pa_to_pvh(pa);
@@ -2081,7 +2043,7 @@ pmap_changebit(pa, bit, setem)
* Clear saved attributes (modify, reference)
*/
if (!setem)
- pmap_attributes[pa_index(pa)] &= ~bit;
+ *pa_to_attribute(pa) &= ~bit;
/*
* Loop over all current mappings setting/clearing as appropos
* If setting RO do we need to clear the VAC?
@@ -2377,18 +2339,21 @@ pmap_enter_ptpage(pmap, va)
splx(s);
}
-static int
-pmap_isvalidphys(pa)
-vm_offset_t pa;
+/*
+ * Routine: pmap_virtual_space
+ *
+ * Function:
+ * Report the range of available kernel virtual address
+ * space to the VM system during bootstrap. Called by
+ * vm_bootstrap_steal_memory().
+ */
+void
+pmap_virtual_space(vstartp, vendp)
+ vm_offset_t *vstartp, *vendp;
{
- struct memseg *sep = &usable_segs[0];
- while (sep->start) {
- if (pa >= sep->start && pa < sep->end)
- return(TRUE);
- ++sep;
- }
- return(FALSE);
+ *vstartp = virtual_avail;
+ *vendp = virtual_end;
}
#ifdef DEBUG
diff --git a/sys/arch/atari/conf/std.atari b/sys/arch/atari/conf/std.atari
index 8fb02eedc88..64853b6a6f5 100644
--- a/sys/arch/atari/conf/std.atari
+++ b/sys/arch/atari/conf/std.atari
@@ -1,10 +1,10 @@
-# $NetBSD: std.atari,v 1.11 1997/11/17 01:57:40 lukem Exp $
+# $NetBSD: std.atari,v 1.12 1998/05/07 07:26:05 leo Exp $
#
# standard atari information
#
machine atari m68k
-options MACHINE_NONCONTIG # Non-contiguous memory support
+options MACHINE_NEW_NONCONTIG # old interface isn't supported anymore
# (required)
options EXEC_SCRIPT # Support #!
diff --git a/sys/arch/atari/include/pmap.h b/sys/arch/atari/include/pmap.h
index 476521df995..68cd4e25ced 100644
--- a/sys/arch/atari/include/pmap.h
+++ b/sys/arch/atari/include/pmap.h
@@ -1,4 +1,4 @@
-/* $NetBSD: pmap.h,v 1.17 1998/02/18 02:05:33 cgd Exp $ */
+/* $NetBSD: pmap.h,v 1.18 1998/05/07 07:26:05 leo Exp $ */
/*
* Copyright (c) 1987 Carnegie-Mellon University
@@ -149,13 +149,6 @@ u_int *Sysmap;
char *vmmap; /* map for mem, dumps, etc. */
struct pmap kernel_pmap_store;
-#ifdef MACHINE_NONCONTIG
-#define pa_index(pa) pmap_page_index(pa)
-#else
-#define pa_index(pa) atop(pa - vm_first_phys)
-#endif /* MACHINE_NONCONTIG */
-
-#define pa_to_pvh(pa) (&pv_table[pa_index(pa)])
#define pmap_kernel() (&kernel_pmap_store)
#define pmap_resident_count(pmap) ((pmap)->pm_stats.resident_count)
diff --git a/sys/arch/atari/include/vmparam.h b/sys/arch/atari/include/vmparam.h
index f22306aa3f1..1a1858242dc 100644
--- a/sys/arch/atari/include/vmparam.h
+++ b/sys/arch/atari/include/vmparam.h
@@ -1,4 +1,4 @@
-/* $NetBSD: vmparam.h,v 1.4 1997/10/10 11:05:53 leo Exp $ */
+/* $NetBSD: vmparam.h,v 1.5 1998/05/07 07:26:05 leo Exp $ */
/*
* Copyright (c) 1988 University of Utah.
@@ -156,6 +156,22 @@
#define VM_PHYS_SIZE (USRIOSIZE*CLBYTES)
/*
+ * Our bootloader currently passes up to 2 segments (ST and TT ram).
+ */
+#define VM_PHYSSEG_MAX (2)
+#define VM_PHYSSEG_STRAT VM_PSTRAT_RANDOM
+#define VM_PHYSSEG_NOADD
+
+/*
+ * pmap-specific data stored in the vm_physmem[] array.
+ */
+struct pmap_physseg {
+ struct pv_entry *pvent; /* pv table for this seg */
+ char *attrs; /* page attributes for this seg */
+};
+
+
+/*
* number of kernel PT pages (initial only, can grow dynamically)
*/
#define VM_KERNEL_PT_PAGES ((vm_size_t)2) /* XXX: SYSPTSIZE */