xref: /freebsd/sys/powerpc/aim/moea64_native.c (revision a25896ca1270e25b657ceaa8d47d5699515f5c25)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD AND 4-Clause-BSD
3  *
4  * Copyright (c) 2001 The NetBSD Foundation, Inc.
5  * All rights reserved.
6  *
7  * This code is derived from software contributed to The NetBSD Foundation
8  * by Matt Thomas <matt@3am-software.com> of Allegro Networks, Inc.
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, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29  * POSSIBILITY OF SUCH DAMAGE.
30  */
31 /*-
32  * Copyright (C) 1995, 1996 Wolfgang Solfrank.
33  * Copyright (C) 1995, 1996 TooLs GmbH.
34  * All rights reserved.
35  *
36  * Redistribution and use in source and binary forms, with or without
37  * modification, are permitted provided that the following conditions
38  * are met:
39  * 1. Redistributions of source code must retain the above copyright
40  *    notice, this list of conditions and the following disclaimer.
41  * 2. Redistributions in binary form must reproduce the above copyright
42  *    notice, this list of conditions and the following disclaimer in the
43  *    documentation and/or other materials provided with the distribution.
44  * 3. All advertising materials mentioning features or use of this software
45  *    must display the following acknowledgement:
46  *	This product includes software developed by TooLs GmbH.
47  * 4. The name of TooLs GmbH may not be used to endorse or promote products
48  *    derived from this software without specific prior written permission.
49  *
50  * THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``AS IS'' AND ANY EXPRESS OR
51  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
52  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
53  * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
54  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
55  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
56  * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
57  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
58  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
59  * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
60  *
61  * $NetBSD: pmap.c,v 1.28 2000/03/26 20:42:36 kleink Exp $
62  */
63 /*-
64  * Copyright (C) 2001 Benno Rice.
65  * All rights reserved.
66  *
67  * Redistribution and use in source and binary forms, with or without
68  * modification, are permitted provided that the following conditions
69  * are met:
70  * 1. Redistributions of source code must retain the above copyright
71  *    notice, this list of conditions and the following disclaimer.
72  * 2. Redistributions in binary form must reproduce the above copyright
73  *    notice, this list of conditions and the following disclaimer in the
74  *    documentation and/or other materials provided with the distribution.
75  *
76  * THIS SOFTWARE IS PROVIDED BY Benno Rice ``AS IS'' AND ANY EXPRESS OR
77  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
78  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
79  * IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
80  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
81  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
82  * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
83  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
84  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
85  * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
86  */
87 
88 #include <sys/cdefs.h>
89 __FBSDID("$FreeBSD$");
90 
91 /*
92  * Native 64-bit page table operations for running without a hypervisor.
93  */
94 
95 #include <sys/param.h>
96 #include <sys/kernel.h>
97 #include <sys/ktr.h>
98 #include <sys/lock.h>
99 #include <sys/mutex.h>
100 #include <sys/proc.h>
101 #include <sys/sched.h>
102 #include <sys/sysctl.h>
103 #include <sys/systm.h>
104 #include <sys/rwlock.h>
105 #include <sys/endian.h>
106 
107 #include <sys/kdb.h>
108 
109 #include <vm/vm.h>
110 #include <vm/vm_param.h>
111 #include <vm/vm_kern.h>
112 #include <vm/vm_page.h>
113 #include <vm/vm_map.h>
114 #include <vm/vm_object.h>
115 #include <vm/vm_extern.h>
116 #include <vm/vm_pageout.h>
117 
118 #include <machine/cpu.h>
119 #include <machine/hid.h>
120 #include <machine/md_var.h>
121 #include <machine/mmuvar.h>
122 
123 #include "mmu_oea64.h"
124 #include "mmu_if.h"
125 #include "moea64_if.h"
126 
127 #define	PTESYNC()	__asm __volatile("ptesync");
128 #define	TLBSYNC()	__asm __volatile("tlbsync; ptesync");
129 #define	SYNC()		__asm __volatile("sync");
130 #define	EIEIO()		__asm __volatile("eieio");
131 
132 #define	VSID_HASH_MASK	0x0000007fffffffffULL
133 
134 /* POWER9 only permits a 64k partition table size. */
135 #define	PART_SIZE	0x10000
136 
137 static int moea64_crop_tlbie;
138 
139 static __inline void
140 TLBIE(uint64_t vpn) {
141 #ifndef __powerpc64__
142 	register_t vpn_hi, vpn_lo;
143 	register_t msr;
144 	register_t scratch, intr;
145 #endif
146 
147 	static volatile u_int tlbie_lock = 0;
148 
149 	vpn <<= ADDR_PIDX_SHFT;
150 
151 	/* Hobo spinlock: we need stronger guarantees than mutexes provide */
152 	while (!atomic_cmpset_int(&tlbie_lock, 0, 1));
153 	isync(); /* Flush instruction queue once lock acquired */
154 
155 	if (moea64_crop_tlbie)
156 		vpn &= ~(0xffffULL << 48);
157 
158 #ifdef __powerpc64__
159 	__asm __volatile("tlbie %0" :: "r"(vpn) : "memory");
160 	__asm __volatile("eieio; tlbsync; ptesync" ::: "memory");
161 #else
162 	vpn_hi = (uint32_t)(vpn >> 32);
163 	vpn_lo = (uint32_t)vpn;
164 
165 	intr = intr_disable();
166 	__asm __volatile("\
167 	    mfmsr %0; \
168 	    mr %1, %0; \
169 	    insrdi %1,%5,1,0; \
170 	    mtmsrd %1; isync; \
171 	    \
172 	    sld %1,%2,%4; \
173 	    or %1,%1,%3; \
174 	    tlbie %1; \
175 	    \
176 	    mtmsrd %0; isync; \
177 	    eieio; \
178 	    tlbsync; \
179 	    ptesync;"
180 	: "=r"(msr), "=r"(scratch) : "r"(vpn_hi), "r"(vpn_lo), "r"(32), "r"(1)
181 	    : "memory");
182 	intr_restore(intr);
183 #endif
184 
185 	/* No barriers or special ops -- taken care of by ptesync above */
186 	tlbie_lock = 0;
187 }
188 
189 #define DISABLE_TRANS(msr)	msr = mfmsr(); mtmsr(msr & ~PSL_DR)
190 #define ENABLE_TRANS(msr)	mtmsr(msr)
191 
192 /*
193  * PTEG data.
194  */
195 static volatile struct lpte *moea64_pteg_table;
196 static struct rwlock moea64_eviction_lock;
197 
198 /*
199  * PTE calls.
200  */
201 static int	moea64_pte_insert_native(mmu_t, struct pvo_entry *);
202 static int64_t	moea64_pte_synch_native(mmu_t, struct pvo_entry *);
203 static int64_t	moea64_pte_clear_native(mmu_t, struct pvo_entry *, uint64_t);
204 static int64_t	moea64_pte_replace_native(mmu_t, struct pvo_entry *, int);
205 static int64_t	moea64_pte_unset_native(mmu_t mmu, struct pvo_entry *);
206 
207 /*
208  * Utility routines.
209  */
210 static void	moea64_bootstrap_native(mmu_t mmup,
211 		    vm_offset_t kernelstart, vm_offset_t kernelend);
212 static void	moea64_cpu_bootstrap_native(mmu_t, int ap);
213 static void	tlbia(void);
214 
215 static mmu_method_t moea64_native_methods[] = {
216 	/* Internal interfaces */
217 	MMUMETHOD(mmu_bootstrap,	moea64_bootstrap_native),
218 	MMUMETHOD(mmu_cpu_bootstrap,	moea64_cpu_bootstrap_native),
219 
220 	MMUMETHOD(moea64_pte_synch,	moea64_pte_synch_native),
221 	MMUMETHOD(moea64_pte_clear,	moea64_pte_clear_native),
222 	MMUMETHOD(moea64_pte_unset,	moea64_pte_unset_native),
223 	MMUMETHOD(moea64_pte_replace,	moea64_pte_replace_native),
224 	MMUMETHOD(moea64_pte_insert,	moea64_pte_insert_native),
225 
226 	{ 0, 0 }
227 };
228 
229 MMU_DEF_INHERIT(oea64_mmu_native, MMU_TYPE_G5, moea64_native_methods,
230     0, oea64_mmu);
231 
232 static int64_t
233 moea64_pte_synch_native(mmu_t mmu, struct pvo_entry *pvo)
234 {
235 	volatile struct lpte *pt = moea64_pteg_table + pvo->pvo_pte.slot;
236 	struct lpte properpt;
237 	uint64_t ptelo;
238 
239 	PMAP_LOCK_ASSERT(pvo->pvo_pmap, MA_OWNED);
240 
241 	moea64_pte_from_pvo(pvo, &properpt);
242 
243 	rw_rlock(&moea64_eviction_lock);
244 	if ((be64toh(pt->pte_hi) & LPTE_AVPN_MASK) !=
245 	    (properpt.pte_hi & LPTE_AVPN_MASK)) {
246 		/* Evicted */
247 		rw_runlock(&moea64_eviction_lock);
248 		return (-1);
249 	}
250 
251 	PTESYNC();
252 	ptelo = be64toh(pt->pte_lo);
253 
254 	rw_runlock(&moea64_eviction_lock);
255 
256 	return (ptelo & (LPTE_REF | LPTE_CHG));
257 }
258 
259 static int64_t
260 moea64_pte_clear_native(mmu_t mmu, struct pvo_entry *pvo, uint64_t ptebit)
261 {
262 	volatile struct lpte *pt = moea64_pteg_table + pvo->pvo_pte.slot;
263 	struct lpte properpt;
264 	uint64_t ptelo;
265 
266 	PMAP_LOCK_ASSERT(pvo->pvo_pmap, MA_OWNED);
267 
268 	moea64_pte_from_pvo(pvo, &properpt);
269 
270 	rw_rlock(&moea64_eviction_lock);
271 	if ((be64toh(pt->pte_hi) & LPTE_AVPN_MASK) !=
272 	    (properpt.pte_hi & LPTE_AVPN_MASK)) {
273 		/* Evicted */
274 		rw_runlock(&moea64_eviction_lock);
275 		return (-1);
276 	}
277 
278 	if (ptebit == LPTE_REF) {
279 		/* See "Resetting the Reference Bit" in arch manual */
280 		PTESYNC();
281 		/* 2-step here safe: precision is not guaranteed */
282 		ptelo = be64toh(pt->pte_lo);
283 
284 		/* One-byte store to avoid touching the C bit */
285 		((volatile uint8_t *)(&pt->pte_lo))[6] =
286 #if BYTE_ORDER == BIG_ENDIAN
287 		    ((uint8_t *)(&properpt.pte_lo))[6];
288 #else
289 		    ((uint8_t *)(&properpt.pte_lo))[1];
290 #endif
291 		rw_runlock(&moea64_eviction_lock);
292 
293 		critical_enter();
294 		TLBIE(pvo->pvo_vpn);
295 		critical_exit();
296 	} else {
297 		rw_runlock(&moea64_eviction_lock);
298 		ptelo = moea64_pte_unset_native(mmu, pvo);
299 		moea64_pte_insert_native(mmu, pvo);
300 	}
301 
302 	return (ptelo & (LPTE_REF | LPTE_CHG));
303 }
304 
305 static int64_t
306 moea64_pte_unset_native(mmu_t mmu, struct pvo_entry *pvo)
307 {
308 	volatile struct lpte *pt = moea64_pteg_table + pvo->pvo_pte.slot;
309 	struct lpte properpt;
310 	uint64_t ptelo;
311 
312 	moea64_pte_from_pvo(pvo, &properpt);
313 
314 	rw_rlock(&moea64_eviction_lock);
315 	if ((be64toh(pt->pte_hi & LPTE_AVPN_MASK)) !=
316 	    (properpt.pte_hi & LPTE_AVPN_MASK)) {
317 		/* Evicted */
318 		moea64_pte_overflow--;
319 		rw_runlock(&moea64_eviction_lock);
320 		return (-1);
321 	}
322 
323 	/*
324 	 * Invalidate the pte, briefly locking it to collect RC bits. No
325 	 * atomics needed since this is protected against eviction by the lock.
326 	 */
327 	isync();
328 	critical_enter();
329 	pt->pte_hi = be64toh((pt->pte_hi & ~LPTE_VALID) | LPTE_LOCKED);
330 	PTESYNC();
331 	TLBIE(pvo->pvo_vpn);
332 	ptelo = be64toh(pt->pte_lo);
333 	*((volatile int32_t *)(&pt->pte_hi) + 1) = 0; /* Release lock */
334 	critical_exit();
335 	rw_runlock(&moea64_eviction_lock);
336 
337 	/* Keep statistics */
338 	moea64_pte_valid--;
339 
340 	return (ptelo & (LPTE_CHG | LPTE_REF));
341 }
342 
343 static int64_t
344 moea64_pte_replace_native(mmu_t mmu, struct pvo_entry *pvo, int flags)
345 {
346 	volatile struct lpte *pt = moea64_pteg_table + pvo->pvo_pte.slot;
347 	struct lpte properpt;
348 	int64_t ptelo;
349 
350 	if (flags == 0) {
351 		/* Just some software bits changing. */
352 		moea64_pte_from_pvo(pvo, &properpt);
353 
354 		rw_rlock(&moea64_eviction_lock);
355 		if ((be64toh(pt->pte_hi) & LPTE_AVPN_MASK) !=
356 		    (properpt.pte_hi & LPTE_AVPN_MASK)) {
357 			rw_runlock(&moea64_eviction_lock);
358 			return (-1);
359 		}
360 		pt->pte_hi = htobe64(properpt.pte_hi);
361 		ptelo = be64toh(pt->pte_lo);
362 		rw_runlock(&moea64_eviction_lock);
363 	} else {
364 		/* Otherwise, need reinsertion and deletion */
365 		ptelo = moea64_pte_unset_native(mmu, pvo);
366 		moea64_pte_insert_native(mmu, pvo);
367 	}
368 
369 	return (ptelo);
370 }
371 
372 static void
373 moea64_cpu_bootstrap_native(mmu_t mmup, int ap)
374 {
375 	int i = 0;
376 	#ifdef __powerpc64__
377 	struct slb *slb = PCPU_GET(aim.slb);
378 	register_t seg0;
379 	#endif
380 
381 	/*
382 	 * Initialize segment registers and MMU
383 	 */
384 
385 	mtmsr(mfmsr() & ~PSL_DR & ~PSL_IR);
386 
387 	/*
388 	 * Install kernel SLB entries
389 	 */
390 
391 	#ifdef __powerpc64__
392 		__asm __volatile ("slbia");
393 		__asm __volatile ("slbmfee %0,%1; slbie %0;" : "=r"(seg0) :
394 		    "r"(0));
395 
396 		for (i = 0; i < n_slbs; i++) {
397 			if (!(slb[i].slbe & SLBE_VALID))
398 				continue;
399 
400 			__asm __volatile ("slbmte %0, %1" ::
401 			    "r"(slb[i].slbv), "r"(slb[i].slbe));
402 		}
403 	#else
404 		for (i = 0; i < 16; i++)
405 			mtsrin(i << ADDR_SR_SHFT, kernel_pmap->pm_sr[i]);
406 	#endif
407 
408 	/*
409 	 * Install page table
410 	 */
411 
412 	__asm __volatile ("ptesync; mtsdr1 %0; isync"
413 	    :: "r"(((uintptr_t)moea64_pteg_table & ~DMAP_BASE_ADDRESS)
414 		     | (uintptr_t)(flsl(moea64_pteg_mask >> 11))));
415 	tlbia();
416 }
417 
418 static void
419 moea64_bootstrap_native(mmu_t mmup, vm_offset_t kernelstart,
420     vm_offset_t kernelend)
421 {
422 	vm_size_t	size;
423 	vm_offset_t	off;
424 	vm_paddr_t	pa;
425 	register_t	msr;
426 
427 	moea64_early_bootstrap(mmup, kernelstart, kernelend);
428 
429 	switch (mfpvr() >> 16) {
430 	case IBMPOWER4:
431 	case IBMPOWER4PLUS:
432 	case IBM970:
433 	case IBM970FX:
434 	case IBM970GX:
435 	case IBM970MP:
436 	    	moea64_crop_tlbie = true;
437 	}
438 	/*
439 	 * Allocate PTEG table.
440 	 */
441 
442 	size = moea64_pteg_count * sizeof(struct lpteg);
443 	CTR2(KTR_PMAP, "moea64_bootstrap: %lu PTEGs, %lu bytes",
444 	    moea64_pteg_count, size);
445 	rw_init(&moea64_eviction_lock, "pte eviction");
446 
447 	/*
448 	 * We now need to allocate memory. This memory, to be allocated,
449 	 * has to reside in a page table. The page table we are about to
450 	 * allocate. We don't have BAT. So drop to data real mode for a minute
451 	 * as a measure of last resort. We do this a couple times.
452 	 */
453 	/*
454 	 * PTEG table must be aligned on a 256k boundary, but can be placed
455 	 * anywhere with that alignment on POWER ISA 3+ systems. On earlier
456 	 * systems, offset addition is done by the CPU with bitwise OR rather
457 	 * than addition, so the table must also be aligned on a boundary of
458 	 * its own size. Pick the larger of the two, which works on all
459 	 * systems.
460 	 */
461 	moea64_pteg_table = (struct lpte *)moea64_bootstrap_alloc(size,
462 	    MAX(256*1024, size));
463 	if (hw_direct_map)
464 		moea64_pteg_table =
465 		    (struct lpte *)PHYS_TO_DMAP((vm_offset_t)moea64_pteg_table);
466 	DISABLE_TRANS(msr);
467 	bzero(__DEVOLATILE(void *, moea64_pteg_table), moea64_pteg_count *
468 	    sizeof(struct lpteg));
469 	ENABLE_TRANS(msr);
470 
471 	CTR1(KTR_PMAP, "moea64_bootstrap: PTEG table at %p", moea64_pteg_table);
472 
473 	moea64_mid_bootstrap(mmup, kernelstart, kernelend);
474 
475 	/*
476 	 * Add a mapping for the page table itself if there is no direct map.
477 	 */
478 	if (!hw_direct_map) {
479 		size = moea64_pteg_count * sizeof(struct lpteg);
480 		off = (vm_offset_t)(moea64_pteg_table);
481 		DISABLE_TRANS(msr);
482 		for (pa = off; pa < off + size; pa += PAGE_SIZE)
483 			pmap_kenter(pa, pa);
484 		ENABLE_TRANS(msr);
485 	}
486 
487 	/* Bring up virtual memory */
488 	moea64_late_bootstrap(mmup, kernelstart, kernelend);
489 }
490 
491 static void
492 tlbia(void)
493 {
494 	vm_offset_t i;
495 	#ifndef __powerpc64__
496 	register_t msr, scratch;
497 	#endif
498 
499 	i = 0xc00; /* IS = 11 */
500 	switch (mfpvr() >> 16) {
501 	case IBM970:
502 	case IBM970FX:
503 	case IBM970MP:
504 	case IBM970GX:
505 	case IBMPOWER4:
506 	case IBMPOWER4PLUS:
507 	case IBMPOWER5:
508 	case IBMPOWER5PLUS:
509 		i = 0; /* IS not supported */
510 		break;
511 	}
512 
513 	TLBSYNC();
514 
515 	for (; i < 0x200000; i += 0x00001000) {
516 		#ifdef __powerpc64__
517 		__asm __volatile("tlbiel %0" :: "r"(i));
518 		#else
519 		__asm __volatile("\
520 		    mfmsr %0; \
521 		    mr %1, %0; \
522 		    insrdi %1,%3,1,0; \
523 		    mtmsrd %1; \
524 		    isync; \
525 		    \
526 		    tlbiel %2; \
527 		    \
528 		    mtmsrd %0; \
529 		    isync;"
530 		: "=r"(msr), "=r"(scratch) : "r"(i), "r"(1));
531 		#endif
532 	}
533 
534 	EIEIO();
535 	TLBSYNC();
536 }
537 
538 static int
539 atomic_pte_lock(volatile struct lpte *pte, uint64_t bitmask, uint64_t *oldhi)
540 {
541 	int	ret;
542 	uint32_t oldhihalf;
543 
544 	/*
545 	 * Note: in principle, if just the locked bit were set here, we
546 	 * could avoid needing the eviction lock. However, eviction occurs
547 	 * so rarely that it isn't worth bothering about in practice.
548 	 */
549 
550 	__asm __volatile (
551 		"1:\tlwarx %1, 0, %3\n\t"	/* load old value */
552 		"and. %0,%1,%4\n\t"		/* check if any bits set */
553 		"bne 2f\n\t"			/* exit if any set */
554 		"stwcx. %5, 0, %3\n\t"      	/* attempt to store */
555 		"bne- 1b\n\t"			/* spin if failed */
556 		"li %0, 1\n\t"			/* success - retval = 1 */
557 		"b 3f\n\t"			/* we've succeeded */
558 		"2:\n\t"
559 		"stwcx. %1, 0, %3\n\t"       	/* clear reservation (74xx) */
560 		"li %0, 0\n\t"			/* failure - retval = 0 */
561 		"3:\n\t"
562 		: "=&r" (ret), "=&r"(oldhihalf), "=m" (pte->pte_hi)
563 		: "r" ((volatile char *)&pte->pte_hi + 4),
564 		  "r" ((uint32_t)bitmask), "r" ((uint32_t)LPTE_LOCKED),
565 		  "m" (pte->pte_hi)
566 		: "cr0", "cr1", "cr2", "memory");
567 
568 	*oldhi = (pte->pte_hi & 0xffffffff00000000ULL) | oldhihalf;
569 
570 	return (ret);
571 }
572 
573 static uintptr_t
574 moea64_insert_to_pteg_native(struct lpte *pvo_pt, uintptr_t slotbase,
575     uint64_t mask)
576 {
577 	volatile struct lpte *pt;
578 	uint64_t oldptehi, va;
579 	uintptr_t k;
580 	int i, j;
581 
582 	/* Start at a random slot */
583 	i = mftb() % 8;
584 	for (j = 0; j < 8; j++) {
585 		k = slotbase + (i + j) % 8;
586 		pt = &moea64_pteg_table[k];
587 		/* Invalidate and seize lock only if no bits in mask set */
588 		if (atomic_pte_lock(pt, mask, &oldptehi)) /* Lock obtained */
589 			break;
590 	}
591 
592 	if (j == 8)
593 		return (-1);
594 
595 	if (oldptehi & LPTE_VALID) {
596 		KASSERT(!(oldptehi & LPTE_WIRED), ("Unmapped wired entry"));
597 		/*
598 		 * Need to invalidate old entry completely: see
599 		 * "Modifying a Page Table Entry". Need to reconstruct
600 		 * the virtual address for the outgoing entry to do that.
601 		 */
602 		if (oldptehi & LPTE_BIG)
603 			va = oldptehi >> moea64_large_page_shift;
604 		else
605 			va = oldptehi >> ADDR_PIDX_SHFT;
606 		if (oldptehi & LPTE_HID)
607 			va = (((k >> 3) ^ moea64_pteg_mask) ^ va) &
608 			    VSID_HASH_MASK;
609 		else
610 			va = ((k >> 3) ^ va) & VSID_HASH_MASK;
611 		va |= (oldptehi & LPTE_AVPN_MASK) <<
612 		    (ADDR_API_SHFT64 - ADDR_PIDX_SHFT);
613 		PTESYNC();
614 		TLBIE(va);
615 		moea64_pte_valid--;
616 		moea64_pte_overflow++;
617 	}
618 
619 	/*
620 	 * Update the PTE as per "Adding a Page Table Entry". Lock is released
621 	 * by setting the high doubleworld.
622 	 */
623 	pt->pte_lo = htobe64(pvo_pt->pte_lo);
624 	EIEIO();
625 	pt->pte_hi = htobe64(pvo_pt->pte_hi);
626 	PTESYNC();
627 
628 	/* Keep statistics */
629 	moea64_pte_valid++;
630 
631 	return (k);
632 }
633 
634 static int
635 moea64_pte_insert_native(mmu_t mmu, struct pvo_entry *pvo)
636 {
637 	struct lpte insertpt;
638 	uintptr_t slot;
639 
640 	/* Initialize PTE */
641 	moea64_pte_from_pvo(pvo, &insertpt);
642 
643 	/* Make sure further insertion is locked out during evictions */
644 	rw_rlock(&moea64_eviction_lock);
645 
646 	/*
647 	 * First try primary hash.
648 	 */
649 	pvo->pvo_pte.slot &= ~7ULL; /* Base slot address */
650 	slot = moea64_insert_to_pteg_native(&insertpt, pvo->pvo_pte.slot,
651 	    LPTE_VALID | LPTE_WIRED | LPTE_LOCKED);
652 	if (slot != -1) {
653 		rw_runlock(&moea64_eviction_lock);
654 		pvo->pvo_pte.slot = slot;
655 		return (0);
656 	}
657 
658 	/*
659 	 * Now try secondary hash.
660 	 */
661 	pvo->pvo_vaddr ^= PVO_HID;
662 	insertpt.pte_hi ^= LPTE_HID;
663 	pvo->pvo_pte.slot ^= (moea64_pteg_mask << 3);
664 	slot = moea64_insert_to_pteg_native(&insertpt, pvo->pvo_pte.slot,
665 	    LPTE_VALID | LPTE_WIRED | LPTE_LOCKED);
666 	if (slot != -1) {
667 		rw_runlock(&moea64_eviction_lock);
668 		pvo->pvo_pte.slot = slot;
669 		return (0);
670 	}
671 
672 	/*
673 	 * Out of luck. Find a PTE to sacrifice.
674 	 */
675 
676 	/* Lock out all insertions for a bit */
677 	if (!rw_try_upgrade(&moea64_eviction_lock)) {
678 		rw_runlock(&moea64_eviction_lock);
679 		rw_wlock(&moea64_eviction_lock);
680 	}
681 
682 	slot = moea64_insert_to_pteg_native(&insertpt, pvo->pvo_pte.slot,
683 	    LPTE_WIRED | LPTE_LOCKED);
684 	if (slot != -1) {
685 		rw_wunlock(&moea64_eviction_lock);
686 		pvo->pvo_pte.slot = slot;
687 		return (0);
688 	}
689 
690 	/* Try other hash table. Now we're getting desperate... */
691 	pvo->pvo_vaddr ^= PVO_HID;
692 	insertpt.pte_hi ^= LPTE_HID;
693 	pvo->pvo_pte.slot ^= (moea64_pteg_mask << 3);
694 	slot = moea64_insert_to_pteg_native(&insertpt, pvo->pvo_pte.slot,
695 	    LPTE_WIRED | LPTE_LOCKED);
696 	if (slot != -1) {
697 		rw_wunlock(&moea64_eviction_lock);
698 		pvo->pvo_pte.slot = slot;
699 		return (0);
700 	}
701 
702 	/* No freeable slots in either PTEG? We're hosed. */
703 	rw_wunlock(&moea64_eviction_lock);
704 	panic("moea64_pte_insert: overflow");
705 	return (-1);
706 }
707 
708