PGMPool.cpp@ 97178

Last change on this file since 97178 was 97145, checked in by vboxsync, 2 years ago
VMM/PGM: Nested VMX: bugref:10092 Implemented large (2M) page support for EPT PDEs.
Property svn:eol-style set to `native` Property svn:keywords set to `Id Revision`
File size: 74.3 KB

Line
1	/* $Id: PGMPool.cpp 97145 2022-10-14 06:05:15Z vboxsync $ */
2	/** @file
3	* PGM Shadow Page Pool.
4	*/
5
6	/*
7	* Copyright (C) 2006-2022 Oracle and/or its affiliates.
8	*
9	* This file is part of VirtualBox base platform packages, as
10	* available from https://www.virtualbox.org.
11	*
12	* This program is free software; you can redistribute it and/or
13	* modify it under the terms of the GNU General Public License
14	* as published by the Free Software Foundation, in version 3 of the
15	* License.
16	*
17	* This program is distributed in the hope that it will be useful, but
18	* WITHOUT ANY WARRANTY; without even the implied warranty of
19	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20	* General Public License for more details.
21	*
22	* You should have received a copy of the GNU General Public License
23	* along with this program; if not, see <https://www.gnu.org/licenses>.
24	*
25	* SPDX-License-Identifier: GPL-3.0-only
26	*/
27
28	/** @page pg_pgm_pool PGM Shadow Page Pool
29	*
30	* Motivations:
31	* -# Relationship between shadow page tables and physical guest pages. This
32	* should allow us to skip most of the global flushes now following access
33	* handler changes. The main expense is flushing shadow pages.
34	* -# Limit the pool size if necessary (default is kind of limitless).
35	* -# Allocate shadow pages from RC. We use to only do this in SyncCR3.
36	* -# Required for 64-bit guests.
37	* -# Combining the PD cache and page pool in order to simplify caching.
38	*
39	*
40	* @section sec_pgm_pool_outline Design Outline
41	*
42	* The shadow page pool tracks pages used for shadowing paging structures (i.e.
43	* page tables, page directory, page directory pointer table and page map
44	* level-4). Each page in the pool has an unique identifier. This identifier is
45	* used to link a guest physical page to a shadow PT. The identifier is a
46	* non-zero value and has a relativly low max value - say 14 bits. This makes it
47	* possible to fit it into the upper bits of the of the aHCPhys entries in the
48	* ram range.
49	*
50	* By restricting host physical memory to the first 48 bits (which is the
51	* announced physical memory range of the K8L chip (scheduled for 2008)), we
52	* can safely use the upper 16 bits for shadow page ID and reference counting.
53	*
54	* Update: The 48 bit assumption will be lifted with the new physical memory
55	* management (PGMPAGE), so we won't have any trouble when someone stuffs 2TB
56	* into a box in some years.
57	*
58	* Now, it's possible for a page to be aliased, i.e. mapped by more than one PT
59	* or PD. This is solved by creating a list of physical cross reference extents
60	* when ever this happens. Each node in the list (extent) is can contain 3 page
61	* pool indexes. The list it self is chained using indexes into the paPhysExt
62	* array.
63	*
64	*
65	* @section sec_pgm_pool_life Life Cycle of a Shadow Page
66	*
67	* -# The SyncPT function requests a page from the pool.
68	* The request includes the kind of page it is (PT/PD, PAE/legacy), the
69	* address of the page it's shadowing, and more.
70	* -# The pool responds to the request by allocating a new page.
71	* When the cache is enabled, it will first check if it's in the cache.
72	* Should the pool be exhausted, one of two things can be done:
73	* -# Flush the whole pool and current CR3.
74	* -# Use the cache to find a page which can be flushed (~age).
75	* -# The SyncPT function will sync one or more pages and insert it into the
76	* shadow PD.
77	* -# The SyncPage function may sync more pages on a later \#PFs.
78	* -# The page is freed / flushed in SyncCR3 (perhaps) and some other cases.
79	* When caching is enabled, the page isn't flush but remains in the cache.
80	*
81	*
82	* @section sec_pgm_pool_monitoring Monitoring
83	*
84	* We always monitor GUEST_PAGE_SIZE chunks of memory. When we've got multiple
85	* shadow pages for the same GUEST_PAGE_SIZE of guest memory (PAE and mixed
86	* PD/PT) the pages sharing the monitor get linked using the
87	* iMonitoredNext/Prev. The head page is the pvUser to the access handlers.
88	*
89	*
90	* @section sec_pgm_pool_impl Implementation
91	*
92	* The pool will take pages from the MM page pool. The tracking data
93	* (attributes, bitmaps and so on) are allocated from the hypervisor heap. The
94	* pool content can be accessed both by using the page id and the physical
95	* address (HC). The former is managed by means of an array, the latter by an
96	* offset based AVL tree.
97	*
98	* Flushing of a pool page means that we iterate the content (we know what kind
99	* it is) and updates the link information in the ram range.
100	*
101	* ...
102	*/
103
104
105	/*********************************************************************************************************************************
106	* Header Files *
107	*********************************************************************************************************************************/
108	#define LOG_GROUP LOG_GROUP_PGM_POOL
109	#define VBOX_WITHOUT_PAGING_BIT_FIELDS /* 64-bit bitfields are just asking for trouble. See @bugref{9841} and others. */
110	#include <VBox/vmm/pgm.h>
111	#include <VBox/vmm/mm.h>
112	#include "PGMInternal.h"
113	#include <VBox/vmm/vmcc.h>
114	#include <VBox/vmm/uvm.h>
115	#include "PGMInline.h"
116
117	#include <VBox/log.h>
118	#include <VBox/err.h>
119	#include <iprt/asm.h>
120	#include <iprt/string.h>
121	#include <VBox/dbg.h>
122
123
124	/*********************************************************************************************************************************
125	* Structures and Typedefs *
126	*********************************************************************************************************************************/
127	typedef struct PGMPOOLCHECKERSTATE
128	{
129	PDBGCCMDHLP pCmdHlp;
130	PVM pVM;
131	PPGMPOOL pPool;
132	PPGMPOOLPAGE pPage;
133	bool fFirstMsg;
134	uint32_t cErrors;
135	} PGMPOOLCHECKERSTATE;
136	typedef PGMPOOLCHECKERSTATE *PPGMPOOLCHECKERSTATE;
137
138
139
140	/*********************************************************************************************************************************
141	* Internal Functions *
142	*********************************************************************************************************************************/
143	static FNDBGFHANDLERINT pgmR3PoolInfoPages;
144	static FNDBGFHANDLERINT pgmR3PoolInfoRoots;
145
146	#ifdef VBOX_WITH_DEBUGGER
147	static FNDBGCCMD pgmR3PoolCmdCheck;
148
149	/** Command descriptors. */
150	static const DBGCCMD g_aCmds[] =
151	{
152	/* pszCmd, cArgsMin, cArgsMax, paArgDesc, cArgDescs, fFlags, pfnHandler pszSyntax, ....pszDescription */
153	{ "pgmpoolcheck", 0, 0, NULL, 0, 0, pgmR3PoolCmdCheck, "", "Check the pgm pool pages." },
154	};
155	#endif
156
157	/**
158	* Initializes the pool
159	*
160	* @returns VBox status code.
161	* @param pVM The cross context VM structure.
162	*/
163	int pgmR3PoolInit(PVM pVM)
164	{
165	int rc;
166
167	AssertCompile(NIL_PGMPOOL_IDX == 0);
168	/* pPage->cLocked is an unsigned byte. */
169	AssertCompile(VMM_MAX_CPU_COUNT <= 255);
170
171	/*
172	* Query Pool config.
173	*/
174	PCFGMNODE pCfg = CFGMR3GetChild(CFGMR3GetRoot(pVM), "/PGM/Pool");
175
176	/* Default pgm pool size is 1024 pages (4MB). */
177	uint16_t cMaxPages = 1024;
178
179	/* Adjust it up relative to the RAM size, using the nested paging formula. */
180	uint64_t cbRam;
181	rc = CFGMR3QueryU64Def(CFGMR3GetRoot(pVM), "RamSize", &cbRam, 0); AssertRCReturn(rc, rc);
182	/** @todo guest x86 specific */
183	uint64_t u64MaxPages = (cbRam >> 9)
184	+ (cbRam >> 18)
185	+ (cbRam >> 27)
186	+ 32 * GUEST_PAGE_SIZE;
187	u64MaxPages >>= GUEST_PAGE_SHIFT;
188	if (u64MaxPages > PGMPOOL_IDX_LAST)
189	cMaxPages = PGMPOOL_IDX_LAST;
190	else
191	cMaxPages = (uint16_t)u64MaxPages;
192
193	/** @cfgm{/PGM/Pool/MaxPages, uint16_t, \#pages, 16, 0x3fff, F(ram-size)}
194	* The max size of the shadow page pool in pages. The pool will grow dynamically
195	* up to this limit.
196	*/
197	rc = CFGMR3QueryU16Def(pCfg, "MaxPages", &cMaxPages, cMaxPages);
198	AssertLogRelRCReturn(rc, rc);
199	AssertLogRelMsgReturn(cMaxPages <= PGMPOOL_IDX_LAST && cMaxPages >= RT_ALIGN(PGMPOOL_IDX_FIRST, 16),
200	("cMaxPages=%u (%#x)\n", cMaxPages, cMaxPages), VERR_INVALID_PARAMETER);
201	AssertCompile(RT_IS_POWER_OF_TWO(PGMPOOL_CFG_MAX_GROW));
202	if (cMaxPages < PGMPOOL_IDX_LAST)
203	cMaxPages = RT_ALIGN(cMaxPages, PGMPOOL_CFG_MAX_GROW / 2);
204	if (cMaxPages > PGMPOOL_IDX_LAST)
205	cMaxPages = PGMPOOL_IDX_LAST;
206	LogRel(("PGM: PGMPool: cMaxPages=%u (u64MaxPages=%llu)\n", cMaxPages, u64MaxPages));
207
208	/** @todo
209	* We need to be much more careful with our allocation strategy here.
210	* For nested paging we don't need pool user info nor extents at all, but
211	* we can't check for nested paging here (too early during init to get a
212	* confirmation it can be used). The default for large memory configs is a
213	* bit large for shadow paging, so I've restricted the extent maximum to 8k
214	* (8k * 16 = 128k of hyper heap).
215	*
216	* Also when large page support is enabled, we typically don't need so much,
217	* although that depends on the availability of 2 MB chunks on the host.
218	*/
219
220	/** @cfgm{/PGM/Pool/MaxUsers, uint16_t, \#users, MaxUsers, 32K, MaxPages*2}
221	* The max number of shadow page user tracking records. Each shadow page has
222	* zero of other shadow pages (or CR3s) that references it, or uses it if you
223	* like. The structures describing these relationships are allocated from a
224	* fixed sized pool. This configuration variable defines the pool size.
225	*/
226	uint16_t cMaxUsers;
227	rc = CFGMR3QueryU16Def(pCfg, "MaxUsers", &cMaxUsers, cMaxPages * 2);
228	AssertLogRelRCReturn(rc, rc);
229	AssertLogRelMsgReturn(cMaxUsers >= cMaxPages && cMaxPages <= _32K,
230	("cMaxUsers=%u (%#x)\n", cMaxUsers, cMaxUsers), VERR_INVALID_PARAMETER);
231
232	/** @cfgm{/PGM/Pool/MaxPhysExts, uint16_t, \#extents, 16, MaxPages * 2, MIN(MaxPages*2\,8192)}
233	* The max number of extents for tracking aliased guest pages.
234	*/
235	uint16_t cMaxPhysExts;
236	rc = CFGMR3QueryU16Def(pCfg, "MaxPhysExts", &cMaxPhysExts,
237	RT_MIN(cMaxPages * 2, 8192 /* 8Ki max as this eat too much hyper heap */));
238	AssertLogRelRCReturn(rc, rc);
239	AssertLogRelMsgReturn(cMaxPhysExts >= 16 && cMaxPhysExts <= PGMPOOL_IDX_LAST,
240	("cMaxPhysExts=%u (%#x)\n", cMaxPhysExts, cMaxPhysExts), VERR_INVALID_PARAMETER);
241
242	/** @cfgm{/PGM/Pool/ChacheEnabled, bool, true}
243	* Enables or disabling caching of shadow pages. Caching means that we will try
244	* reuse shadow pages instead of recreating them everything SyncCR3, SyncPT or
245	* SyncPage requests one. When reusing a shadow page, we can save time
246	* reconstructing it and it's children.
247	*/
248	bool fCacheEnabled;
249	rc = CFGMR3QueryBoolDef(pCfg, "CacheEnabled", &fCacheEnabled, true);
250	AssertLogRelRCReturn(rc, rc);
251
252	LogRel(("PGM: pgmR3PoolInit: cMaxPages=%#RX16 cMaxUsers=%#RX16 cMaxPhysExts=%#RX16 fCacheEnable=%RTbool\n",
253	cMaxPages, cMaxUsers, cMaxPhysExts, fCacheEnabled));
254
255	/*
256	* Allocate the data structures.
257	*/
258	uint32_t cb = RT_UOFFSETOF_DYN(PGMPOOL, aPages[cMaxPages]);
259	cb += cMaxUsers * sizeof(PGMPOOLUSER);
260	cb += cMaxPhysExts * sizeof(PGMPOOLPHYSEXT);
261	PPGMPOOL pPool;
262	RTR0PTR pPoolR0;
263	rc = SUPR3PageAllocEx(RT_ALIGN_32(cb, HOST_PAGE_SIZE) >> HOST_PAGE_SHIFT, 0 /fFlags/, (void **)&pPool, &pPoolR0, NULL);
264	if (RT_FAILURE(rc))
265	return rc;
266	Assert(ASMMemIsZero(pPool, cb));
267	pVM->pgm.s.pPoolR3 = pPool->pPoolR3 = pPool;
268	pVM->pgm.s.pPoolR0 = pPool->pPoolR0 = pPoolR0;
269
270	/*
271	* Initialize it.
272	*/
273	pPool->pVMR3 = pVM;
274	pPool->pVMR0 = pVM->pVMR0ForCall;
275	pPool->cMaxPages = cMaxPages;
276	pPool->cCurPages = PGMPOOL_IDX_FIRST;
277	pPool->iUserFreeHead = 0;
278	pPool->cMaxUsers = cMaxUsers;
279	PPGMPOOLUSER paUsers = (PPGMPOOLUSER)&pPool->aPages[pPool->cMaxPages];
280	pPool->paUsersR3 = paUsers;
281	pPool->paUsersR0 = pPoolR0 + (uintptr_t)paUsers - (uintptr_t)pPool;
282	for (unsigned i = 0; i < cMaxUsers; i++)
283	{
284	paUsers[i].iNext = i + 1;
285	paUsers[i].iUser = NIL_PGMPOOL_IDX;
286	paUsers[i].iUserTable = 0xfffffffe;
287	}
288	paUsers[cMaxUsers - 1].iNext = NIL_PGMPOOL_USER_INDEX;
289	pPool->iPhysExtFreeHead = 0;
290	pPool->cMaxPhysExts = cMaxPhysExts;
291	PPGMPOOLPHYSEXT paPhysExts = (PPGMPOOLPHYSEXT)&paUsers[cMaxUsers];
292	pPool->paPhysExtsR3 = paPhysExts;
293	pPool->paPhysExtsR0 = pPoolR0 + (uintptr_t)paPhysExts - (uintptr_t)pPool;
294	for (unsigned i = 0; i < cMaxPhysExts; i++)
295	{
296	paPhysExts[i].iNext = i + 1;
297	paPhysExts[i].aidx[0] = NIL_PGMPOOL_IDX;
298	paPhysExts[i].apte[0] = NIL_PGMPOOL_PHYSEXT_IDX_PTE;
299	paPhysExts[i].aidx[1] = NIL_PGMPOOL_IDX;
300	paPhysExts[i].apte[1] = NIL_PGMPOOL_PHYSEXT_IDX_PTE;
301	paPhysExts[i].aidx[2] = NIL_PGMPOOL_IDX;
302	paPhysExts[i].apte[2] = NIL_PGMPOOL_PHYSEXT_IDX_PTE;
303	}
304	paPhysExts[cMaxPhysExts - 1].iNext = NIL_PGMPOOL_PHYSEXT_INDEX;
305	for (unsigned i = 0; i < RT_ELEMENTS(pPool->aiHash); i++)
306	pPool->aiHash[i] = NIL_PGMPOOL_IDX;
307	pPool->iAgeHead = NIL_PGMPOOL_IDX;
308	pPool->iAgeTail = NIL_PGMPOOL_IDX;
309	pPool->fCacheEnabled = fCacheEnabled;
310
311	pPool->hAccessHandlerType = NIL_PGMPHYSHANDLERTYPE;
312	rc = PGMR3HandlerPhysicalTypeRegister(pVM, PGMPHYSHANDLERKIND_WRITE, PGMPHYSHANDLER_F_KEEP_PGM_LOCK,
313	pgmPoolAccessHandler, "Guest Paging Access Handler", &pPool->hAccessHandlerType);
314	AssertLogRelRCReturn(rc, rc);
315
316	pPool->HCPhysTree = 0;
317
318	/*
319	* The NIL entry.
320	*/
321	Assert(NIL_PGMPOOL_IDX == 0);
322	pPool->aPages[NIL_PGMPOOL_IDX].enmKind = PGMPOOLKIND_INVALID;
323	pPool->aPages[NIL_PGMPOOL_IDX].idx = NIL_PGMPOOL_IDX;
324	pPool->aPages[NIL_PGMPOOL_IDX].Core.Key = NIL_RTHCPHYS;
325	pPool->aPages[NIL_PGMPOOL_IDX].GCPhys = NIL_RTGCPHYS;
326	pPool->aPages[NIL_PGMPOOL_IDX].iNext = NIL_PGMPOOL_IDX;
327	/* pPool->aPages[NIL_PGMPOOL_IDX].cLocked = INT32_MAX; - test this out... */
328	pPool->aPages[NIL_PGMPOOL_IDX].pvPageR3 = 0;
329	pPool->aPages[NIL_PGMPOOL_IDX].iUserHead = NIL_PGMPOOL_USER_INDEX;
330	pPool->aPages[NIL_PGMPOOL_IDX].iModifiedNext = NIL_PGMPOOL_IDX;
331	pPool->aPages[NIL_PGMPOOL_IDX].iModifiedPrev = NIL_PGMPOOL_IDX;
332	pPool->aPages[NIL_PGMPOOL_IDX].iMonitoredNext = NIL_PGMPOOL_IDX;
333	pPool->aPages[NIL_PGMPOOL_IDX].iMonitoredPrev = NIL_PGMPOOL_IDX;
334	pPool->aPages[NIL_PGMPOOL_IDX].iAgeNext = NIL_PGMPOOL_IDX;
335	pPool->aPages[NIL_PGMPOOL_IDX].iAgePrev = NIL_PGMPOOL_IDX;
336
337	Assert(pPool->aPages[NIL_PGMPOOL_IDX].idx == NIL_PGMPOOL_IDX);
338	Assert(pPool->aPages[NIL_PGMPOOL_IDX].GCPhys == NIL_RTGCPHYS);
339	Assert(!pPool->aPages[NIL_PGMPOOL_IDX].fSeenNonGlobal);
340	Assert(!pPool->aPages[NIL_PGMPOOL_IDX].fMonitored);
341	Assert(!pPool->aPages[NIL_PGMPOOL_IDX].fCached);
342	Assert(!pPool->aPages[NIL_PGMPOOL_IDX].fZeroed);
343	Assert(!pPool->aPages[NIL_PGMPOOL_IDX].fReusedFlushPending);
344
345	/*
346	* Register statistics.
347	*/
348	STAM_REL_REG(pVM, &pPool->StatGrow, STAMTYPE_PROFILE, "/PGM/Pool/Grow", STAMUNIT_TICKS_PER_CALL, "Profiling PGMR0PoolGrow");
349	#ifdef VBOX_WITH_STATISTICS
350	STAM_REG(pVM, &pPool->cCurPages, STAMTYPE_U16, "/PGM/Pool/cCurPages", STAMUNIT_PAGES, "Current pool size.");
351	STAM_REG(pVM, &pPool->cMaxPages, STAMTYPE_U16, "/PGM/Pool/cMaxPages", STAMUNIT_PAGES, "Max pool size.");
352	STAM_REG(pVM, &pPool->cUsedPages, STAMTYPE_U16, "/PGM/Pool/cUsedPages", STAMUNIT_PAGES, "The number of pages currently in use.");
353	STAM_REG(pVM, &pPool->cUsedPagesHigh, STAMTYPE_U16_RESET, "/PGM/Pool/cUsedPagesHigh", STAMUNIT_PAGES, "The high watermark for cUsedPages.");
354	STAM_REG(pVM, &pPool->StatAlloc, STAMTYPE_PROFILE_ADV, "/PGM/Pool/Alloc", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmPoolAlloc.");
355	STAM_REG(pVM, &pPool->StatClearAll, STAMTYPE_PROFILE, "/PGM/Pool/ClearAll", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmR3PoolClearAll.");
356	STAM_REG(pVM, &pPool->StatR3Reset, STAMTYPE_PROFILE, "/PGM/Pool/R3Reset", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmR3PoolReset.");
357	STAM_REG(pVM, &pPool->StatFlushPage, STAMTYPE_PROFILE, "/PGM/Pool/FlushPage", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmPoolFlushPage.");
358	STAM_REG(pVM, &pPool->StatFree, STAMTYPE_PROFILE, "/PGM/Pool/Free", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmPoolFree.");
359	STAM_REG(pVM, &pPool->StatForceFlushPage, STAMTYPE_COUNTER, "/PGM/Pool/FlushForce", STAMUNIT_OCCURENCES, "Counting explicit flushes by PGMPoolFlushPage().");
360	STAM_REG(pVM, &pPool->StatForceFlushDirtyPage, STAMTYPE_COUNTER, "/PGM/Pool/FlushForceDirty", STAMUNIT_OCCURENCES, "Counting explicit flushes of dirty pages by PGMPoolFlushPage().");
361	STAM_REG(pVM, &pPool->StatForceFlushReused, STAMTYPE_COUNTER, "/PGM/Pool/FlushReused", STAMUNIT_OCCURENCES, "Counting flushes for reused pages.");
362	STAM_REG(pVM, &pPool->StatZeroPage, STAMTYPE_PROFILE, "/PGM/Pool/ZeroPage", STAMUNIT_TICKS_PER_CALL, "Profiling time spent zeroing pages. Overlaps with Alloc.");
363	STAM_REG(pVM, &pPool->cMaxUsers, STAMTYPE_U16, "/PGM/Pool/Track/cMaxUsers", STAMUNIT_COUNT, "Max user tracking records.");
364	STAM_REG(pVM, &pPool->cPresent, STAMTYPE_U32, "/PGM/Pool/Track/cPresent", STAMUNIT_COUNT, "Number of present page table entries.");
365	STAM_REG(pVM, &pPool->StatTrackDeref, STAMTYPE_PROFILE, "/PGM/Pool/Track/Deref", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmPoolTrackDeref.");
366	STAM_REG(pVM, &pPool->StatTrackFlushGCPhysPT, STAMTYPE_PROFILE, "/PGM/Pool/Track/FlushGCPhysPT", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmPoolTrackFlushGCPhysPT.");
367	STAM_REG(pVM, &pPool->StatTrackFlushGCPhysPTs, STAMTYPE_PROFILE, "/PGM/Pool/Track/FlushGCPhysPTs", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmPoolTrackFlushGCPhysPTs.");
368	STAM_REG(pVM, &pPool->StatTrackFlushGCPhysPTsSlow, STAMTYPE_PROFILE, "/PGM/Pool/Track/FlushGCPhysPTsSlow", STAMUNIT_TICKS_PER_CALL, "Profiling of pgmPoolTrackFlushGCPhysPTsSlow.");
369	STAM_REG(pVM, &pPool->StatTrackFlushEntry, STAMTYPE_COUNTER, "/PGM/Pool/Track/Entry/Flush", STAMUNIT_COUNT, "Nr of flushed entries.");
370	STAM_REG(pVM, &pPool->StatTrackFlushEntryKeep, STAMTYPE_COUNTER, "/PGM/Pool/Track/Entry/Update", STAMUNIT_COUNT, "Nr of updated entries.");
371	STAM_REG(pVM, &pPool->StatTrackFreeUpOneUser, STAMTYPE_COUNTER, "/PGM/Pool/Track/FreeUpOneUser", STAMUNIT_TICKS_PER_CALL, "The number of times we were out of user tracking records.");
372	STAM_REG(pVM, &pPool->StatTrackDerefGCPhys, STAMTYPE_PROFILE, "/PGM/Pool/Track/DrefGCPhys", STAMUNIT_TICKS_PER_CALL, "Profiling deref activity related tracking GC physical pages.");
373	STAM_REG(pVM, &pPool->StatTrackLinearRamSearches, STAMTYPE_COUNTER, "/PGM/Pool/Track/LinearRamSearches", STAMUNIT_OCCURENCES, "The number of times we had to do linear ram searches.");
374	STAM_REG(pVM, &pPool->StamTrackPhysExtAllocFailures,STAMTYPE_COUNTER, "/PGM/Pool/Track/PhysExtAllocFailures", STAMUNIT_OCCURENCES, "The number of failing pgmPoolTrackPhysExtAlloc calls.");
375
376	STAM_REG(pVM, &pPool->StatMonitorPfRZ, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/RZ/#PF", STAMUNIT_TICKS_PER_CALL, "Profiling the RC/R0 #PF access handler.");
377	STAM_REG(pVM, &pPool->StatMonitorPfRZEmulateInstr, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/#PF/EmulateInstr", STAMUNIT_OCCURENCES, "Times we've failed interpreting the instruction.");
378	STAM_REG(pVM, &pPool->StatMonitorPfRZFlushPage, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/RZ/#PF/FlushPage", STAMUNIT_TICKS_PER_CALL, "Profiling the pgmPoolFlushPage calls made from the RC/R0 access handler.");
379	STAM_REG(pVM, &pPool->StatMonitorPfRZFlushReinit, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/#PF/FlushReinit", STAMUNIT_OCCURENCES, "Times we've detected a page table reinit.");
380	STAM_REG(pVM, &pPool->StatMonitorPfRZFlushModOverflow,STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/#PF/FlushOverflow", STAMUNIT_OCCURENCES, "Counting flushes for pages that are modified too often.");
381	STAM_REG(pVM, &pPool->StatMonitorPfRZFork, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/#PF/Fork", STAMUNIT_OCCURENCES, "Times we've detected fork().");
382	STAM_REG(pVM, &pPool->StatMonitorPfRZHandled, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/RZ/#PF/Handled", STAMUNIT_TICKS_PER_CALL, "Profiling the RC/R0 #PF access we've handled (except REP STOSD).");
383	STAM_REG(pVM, &pPool->StatMonitorPfRZIntrFailPatch1, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/#PF/IntrFailPatch1", STAMUNIT_OCCURENCES, "Times we've failed interpreting a patch code instruction.");
384	STAM_REG(pVM, &pPool->StatMonitorPfRZIntrFailPatch2, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/#PF/IntrFailPatch2", STAMUNIT_OCCURENCES, "Times we've failed interpreting a patch code instruction during flushing.");
385	STAM_REG(pVM, &pPool->StatMonitorPfRZRepPrefix, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/#PF/RepPrefix", STAMUNIT_OCCURENCES, "The number of times we've seen rep prefixes we can't handle.");
386	STAM_REG(pVM, &pPool->StatMonitorPfRZRepStosd, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/RZ/#PF/RepStosd", STAMUNIT_TICKS_PER_CALL, "Profiling the REP STOSD cases we've handled.");
387
388	STAM_REG(pVM, &pPool->StatMonitorRZ, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/RZ/IEM", STAMUNIT_TICKS_PER_CALL, "Profiling the regular access handler.");
389	STAM_REG(pVM, &pPool->StatMonitorRZFlushPage, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/RZ/IEM/FlushPage", STAMUNIT_TICKS_PER_CALL, "Profiling the pgmPoolFlushPage calls made from the regular access handler.");
390	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[0], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size01", STAMUNIT_OCCURENCES, "Number of 1 byte accesses.");
391	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[1], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size02", STAMUNIT_OCCURENCES, "Number of 2 byte accesses.");
392	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[2], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size03", STAMUNIT_OCCURENCES, "Number of 3 byte accesses.");
393	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[3], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size04", STAMUNIT_OCCURENCES, "Number of 4 byte accesses.");
394	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[4], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size05", STAMUNIT_OCCURENCES, "Number of 5 byte accesses.");
395	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[5], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size06", STAMUNIT_OCCURENCES, "Number of 6 byte accesses.");
396	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[6], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size07", STAMUNIT_OCCURENCES, "Number of 7 byte accesses.");
397	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[7], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size08", STAMUNIT_OCCURENCES, "Number of 8 byte accesses.");
398	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[8], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size09", STAMUNIT_OCCURENCES, "Number of 9 byte accesses.");
399	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[9], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size0a", STAMUNIT_OCCURENCES, "Number of 10 byte accesses.");
400	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[10], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size0b", STAMUNIT_OCCURENCES, "Number of 11 byte accesses.");
401	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[11], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size0c", STAMUNIT_OCCURENCES, "Number of 12 byte accesses.");
402	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[12], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size0d", STAMUNIT_OCCURENCES, "Number of 13 byte accesses.");
403	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[13], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size0e", STAMUNIT_OCCURENCES, "Number of 14 byte accesses.");
404	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[14], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size0f", STAMUNIT_OCCURENCES, "Number of 15 byte accesses.");
405	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[15], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size10", STAMUNIT_OCCURENCES, "Number of 16 byte accesses.");
406	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[16], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size11-2f", STAMUNIT_OCCURENCES, "Number of 17-31 byte accesses.");
407	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[17], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size20-3f", STAMUNIT_OCCURENCES, "Number of 32-63 byte accesses.");
408	STAM_REG(pVM, &pPool->aStatMonitorRZSizes[18], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Size40+", STAMUNIT_OCCURENCES, "Number of 64+ byte accesses.");
409	STAM_REG(pVM, &pPool->aStatMonitorRZMisaligned[0], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Misaligned1", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 1.");
410	STAM_REG(pVM, &pPool->aStatMonitorRZMisaligned[1], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Misaligned2", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 2.");
411	STAM_REG(pVM, &pPool->aStatMonitorRZMisaligned[2], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Misaligned3", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 3.");
412	STAM_REG(pVM, &pPool->aStatMonitorRZMisaligned[3], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Misaligned4", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 4.");
413	STAM_REG(pVM, &pPool->aStatMonitorRZMisaligned[4], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Misaligned5", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 5.");
414	STAM_REG(pVM, &pPool->aStatMonitorRZMisaligned[5], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Misaligned6", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 6.");
415	STAM_REG(pVM, &pPool->aStatMonitorRZMisaligned[6], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/IEM/Misaligned7", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 7.");
416
417	STAM_REG(pVM, &pPool->StatMonitorRZFaultPT, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/Fault/PT", STAMUNIT_OCCURENCES, "Nr of handled PT faults.");
418	STAM_REG(pVM, &pPool->StatMonitorRZFaultPD, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/Fault/PD", STAMUNIT_OCCURENCES, "Nr of handled PD faults.");
419	STAM_REG(pVM, &pPool->StatMonitorRZFaultPDPT, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/Fault/PDPT", STAMUNIT_OCCURENCES, "Nr of handled PDPT faults.");
420	STAM_REG(pVM, &pPool->StatMonitorRZFaultPML4, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/RZ/Fault/PML4", STAMUNIT_OCCURENCES, "Nr of handled PML4 faults.");
421
422	STAM_REG(pVM, &pPool->StatMonitorR3, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/R3", STAMUNIT_TICKS_PER_CALL, "Profiling the R3 access handler.");
423	STAM_REG(pVM, &pPool->StatMonitorR3FlushPage, STAMTYPE_PROFILE, "/PGM/Pool/Monitor/R3/FlushPage", STAMUNIT_TICKS_PER_CALL, "Profiling the pgmPoolFlushPage calls made from the R3 access handler.");
424	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[0], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size01", STAMUNIT_OCCURENCES, "Number of 1 byte accesses (R3).");
425	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[1], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size02", STAMUNIT_OCCURENCES, "Number of 2 byte accesses (R3).");
426	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[2], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size03", STAMUNIT_OCCURENCES, "Number of 3 byte accesses (R3).");
427	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[3], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size04", STAMUNIT_OCCURENCES, "Number of 4 byte accesses (R3).");
428	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[4], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size05", STAMUNIT_OCCURENCES, "Number of 5 byte accesses (R3).");
429	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[5], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size06", STAMUNIT_OCCURENCES, "Number of 6 byte accesses (R3).");
430	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[6], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size07", STAMUNIT_OCCURENCES, "Number of 7 byte accesses (R3).");
431	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[7], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size08", STAMUNIT_OCCURENCES, "Number of 8 byte accesses (R3).");
432	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[8], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size09", STAMUNIT_OCCURENCES, "Number of 9 byte accesses (R3).");
433	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[9], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size0a", STAMUNIT_OCCURENCES, "Number of 10 byte accesses (R3).");
434	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[10], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size0b", STAMUNIT_OCCURENCES, "Number of 11 byte accesses (R3).");
435	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[11], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size0c", STAMUNIT_OCCURENCES, "Number of 12 byte accesses (R3).");
436	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[12], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size0d", STAMUNIT_OCCURENCES, "Number of 13 byte accesses (R3).");
437	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[13], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size0e", STAMUNIT_OCCURENCES, "Number of 14 byte accesses (R3).");
438	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[14], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size0f", STAMUNIT_OCCURENCES, "Number of 15 byte accesses (R3).");
439	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[15], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size10", STAMUNIT_OCCURENCES, "Number of 16 byte accesses (R3).");
440	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[16], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size11-2f", STAMUNIT_OCCURENCES, "Number of 17-31 byte accesses.");
441	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[17], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size20-3f", STAMUNIT_OCCURENCES, "Number of 32-63 byte accesses.");
442	STAM_REG(pVM, &pPool->aStatMonitorR3Sizes[18], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Size40+", STAMUNIT_OCCURENCES, "Number of 64+ byte accesses.");
443	STAM_REG(pVM, &pPool->aStatMonitorR3Misaligned[0], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Misaligned1", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 1 in R3.");
444	STAM_REG(pVM, &pPool->aStatMonitorR3Misaligned[1], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Misaligned2", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 2 in R3.");
445	STAM_REG(pVM, &pPool->aStatMonitorR3Misaligned[2], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Misaligned3", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 3 in R3.");
446	STAM_REG(pVM, &pPool->aStatMonitorR3Misaligned[3], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Misaligned4", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 4 in R3.");
447	STAM_REG(pVM, &pPool->aStatMonitorR3Misaligned[4], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Misaligned5", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 5 in R3.");
448	STAM_REG(pVM, &pPool->aStatMonitorR3Misaligned[5], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Misaligned6", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 6 in R3.");
449	STAM_REG(pVM, &pPool->aStatMonitorR3Misaligned[6], STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Misaligned7", STAMUNIT_OCCURENCES, "Number of misaligned access with offset 7 in R3.");
450
451	STAM_REG(pVM, &pPool->StatMonitorR3FaultPT, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Fault/PT", STAMUNIT_OCCURENCES, "Nr of handled PT faults.");
452	STAM_REG(pVM, &pPool->StatMonitorR3FaultPD, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Fault/PD", STAMUNIT_OCCURENCES, "Nr of handled PD faults.");
453	STAM_REG(pVM, &pPool->StatMonitorR3FaultPDPT, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Fault/PDPT", STAMUNIT_OCCURENCES, "Nr of handled PDPT faults.");
454	STAM_REG(pVM, &pPool->StatMonitorR3FaultPML4, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/R3/Fault/PML4", STAMUNIT_OCCURENCES, "Nr of handled PML4 faults.");
455
456	STAM_REG(pVM, &pPool->cModifiedPages, STAMTYPE_U16, "/PGM/Pool/Monitor/cModifiedPages", STAMUNIT_PAGES, "The current cModifiedPages value.");
457	STAM_REG(pVM, &pPool->cModifiedPagesHigh, STAMTYPE_U16_RESET, "/PGM/Pool/Monitor/cModifiedPagesHigh", STAMUNIT_PAGES, "The high watermark for cModifiedPages.");
458	STAM_REG(pVM, &pPool->StatResetDirtyPages, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/Dirty/Resets", STAMUNIT_OCCURENCES, "Times we've called pgmPoolResetDirtyPages (and there were dirty page).");
459	STAM_REG(pVM, &pPool->StatDirtyPage, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/Dirty/Pages", STAMUNIT_OCCURENCES, "Times we've called pgmPoolAddDirtyPage.");
460	STAM_REG(pVM, &pPool->StatDirtyPageDupFlush, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/Dirty/FlushDup", STAMUNIT_OCCURENCES, "Times we've had to flush duplicates for dirty page management.");
461	STAM_REG(pVM, &pPool->StatDirtyPageOverFlowFlush, STAMTYPE_COUNTER, "/PGM/Pool/Monitor/Dirty/FlushOverflow",STAMUNIT_OCCURENCES, "Times we've had to flush because of overflow.");
462	STAM_REG(pVM, &pPool->StatCacheHits, STAMTYPE_COUNTER, "/PGM/Pool/Cache/Hits", STAMUNIT_OCCURENCES, "The number of pgmPoolAlloc calls satisfied by the cache.");
463	STAM_REG(pVM, &pPool->StatCacheMisses, STAMTYPE_COUNTER, "/PGM/Pool/Cache/Misses", STAMUNIT_OCCURENCES, "The number of pgmPoolAlloc calls not statisfied by the cache.");
464	STAM_REG(pVM, &pPool->StatCacheKindMismatches, STAMTYPE_COUNTER, "/PGM/Pool/Cache/KindMismatches", STAMUNIT_OCCURENCES, "The number of shadow page kind mismatches. (Better be low, preferably 0!)");
465	STAM_REG(pVM, &pPool->StatCacheFreeUpOne, STAMTYPE_COUNTER, "/PGM/Pool/Cache/FreeUpOne", STAMUNIT_OCCURENCES, "The number of times the cache was asked to free up a page.");
466	STAM_REG(pVM, &pPool->StatCacheCacheable, STAMTYPE_COUNTER, "/PGM/Pool/Cache/Cacheable", STAMUNIT_OCCURENCES, "The number of cacheable allocations.");
467	STAM_REG(pVM, &pPool->StatCacheUncacheable, STAMTYPE_COUNTER, "/PGM/Pool/Cache/Uncacheable", STAMUNIT_OCCURENCES, "The number of uncacheable allocations.");
468	#endif /* VBOX_WITH_STATISTICS */
469
470	DBGFR3InfoRegisterInternalEx(pVM, "pgmpoolpages", "Lists page pool pages.", pgmR3PoolInfoPages, 0);
471	DBGFR3InfoRegisterInternalEx(pVM, "pgmpoolroots", "Lists page pool roots.", pgmR3PoolInfoRoots, 0);
472
473	#ifdef VBOX_WITH_DEBUGGER
474	/*
475	* Debugger commands.
476	*/
477	static bool s_fRegisteredCmds = false;
478	if (!s_fRegisteredCmds)
479	{
480	rc = DBGCRegisterCommands(&g_aCmds[0], RT_ELEMENTS(g_aCmds));
481	if (RT_SUCCESS(rc))
482	s_fRegisteredCmds = true;
483	}
484	#endif
485
486	return VINF_SUCCESS;
487	}
488
489
490	/**
491	* Relocate the page pool data.
492	*
493	* @param pVM The cross context VM structure.
494	*/
495	void pgmR3PoolRelocate(PVM pVM)
496	{
497	RT_NOREF(pVM);
498	}
499
500
501	/**
502	* Grows the shadow page pool.
503	*
504	* I.e. adds more pages to it, assuming that hasn't reached cMaxPages yet.
505	*
506	* @returns VBox status code.
507	* @param pVM The cross context VM structure.
508	* @param pVCpu The cross context virtual CPU structure of the calling EMT.
509	*/
510	VMMR3_INT_DECL(int) PGMR3PoolGrow(PVM pVM, PVMCPU pVCpu)
511	{
512	/* This used to do a lot of stuff, but it has moved to ring-0 (PGMR0PoolGrow). */
513	AssertReturn(pVM->pgm.s.pPoolR3->cCurPages < pVM->pgm.s.pPoolR3->cMaxPages, VERR_PGM_POOL_MAXED_OUT_ALREADY);
514	int rc = VMMR3CallR0Emt(pVM, pVCpu, VMMR0_DO_PGM_POOL_GROW, 0, NULL);
515	if (rc == VINF_SUCCESS)
516	return rc;
517	LogRel(("PGMR3PoolGrow: rc=%Rrc cCurPages=%#x cMaxPages=%#x\n",
518	rc, pVM->pgm.s.pPoolR3->cCurPages, pVM->pgm.s.pPoolR3->cMaxPages));
519	if (pVM->pgm.s.pPoolR3->cCurPages > 128 && RT_FAILURE_NP(rc))
520	return -rc;
521	return rc;
522	}
523
524
525	/**
526	* Rendezvous callback used by pgmR3PoolClearAll that clears all shadow pages
527	* and all modification counters.
528	*
529	* This is only called on one of the EMTs while the other ones are waiting for
530	* it to complete this function.
531	*
532	* @returns VINF_SUCCESS (VBox strict status code).
533	* @param pVM The cross context VM structure.
534	* @param pVCpu The cross context virtual CPU structure of the calling EMT. Unused.
535	* @param fpvFlushRemTlb When not NULL, we'll flush the REM TLB as well.
536	* (This is the pvUser, so it has to be void *.)
537	*
538	*/
539	DECLCALLBACK(VBOXSTRICTRC) pgmR3PoolClearAllRendezvous(PVM pVM, PVMCPU pVCpu, void *fpvFlushRemTlb)
540	{
541	PPGMPOOL pPool = pVM->pgm.s.CTX_SUFF(pPool);
542	STAM_PROFILE_START(&pPool->StatClearAll, c);
543	NOREF(pVCpu);
544
545	PGM_LOCK_VOID(pVM);
546	Log(("pgmR3PoolClearAllRendezvous: cUsedPages=%d fpvFlushRemTlb=%RTbool\n", pPool->cUsedPages, !!fpvFlushRemTlb));
547
548	/*
549	* Iterate all the pages until we've encountered all that are in use.
550	* This is a simple but not quite optimal solution.
551	*/
552	unsigned cModifiedPages = 0; NOREF(cModifiedPages);
553	unsigned cLeft = pPool->cUsedPages;
554	uint32_t iPage = pPool->cCurPages;
555	while (--iPage >= PGMPOOL_IDX_FIRST)
556	{
557	PPGMPOOLPAGE pPage = &pPool->aPages[iPage];
558	if (pPage->GCPhys != NIL_RTGCPHYS)
559	{
560	switch (pPage->enmKind)
561	{
562	/*
563	* We only care about shadow page tables that reference physical memory
564	*/
565	#ifdef PGM_WITH_LARGE_PAGES
566	case PGMPOOLKIND_PAE_PD_PHYS: /* Large pages reference 2 MB of physical memory, so we must clear them. */
567	if (pPage->cPresent)
568	{
569	PX86PDPAE pShwPD = (PX86PDPAE)PGMPOOL_PAGE_2_PTR_V2(pPool->CTX_SUFF(pVM), pVCpu, pPage);
570	for (unsigned i = 0; i < RT_ELEMENTS(pShwPD->a); i++)
571	{
572	//Assert((pShwPD->a[i].u & UINT64_C(0xfff0000000000f80)) == 0); - bogus, includes X86_PDE_PS.
573	if ((pShwPD->a[i].u & (X86_PDE_P \| X86_PDE_PS)) == (X86_PDE_P \| X86_PDE_PS))
574	{
575	pShwPD->a[i].u = 0;
576	Assert(pPage->cPresent);
577	pPage->cPresent--;
578	}
579	}
580	if (pPage->cPresent == 0)
581	pPage->iFirstPresent = NIL_PGMPOOL_PRESENT_INDEX;
582	}
583	goto default_case;
584
585	case PGMPOOLKIND_EPT_PD_FOR_PHYS: /* Large pages reference 2 MB of physical memory, so we must clear them. */
586	if (pPage->cPresent)
587	{
588	PEPTPD pShwPD = (PEPTPD)PGMPOOL_PAGE_2_PTR_V2(pPool->CTX_SUFF(pVM), pVCpu, pPage);
589	for (unsigned i = 0; i < RT_ELEMENTS(pShwPD->a); i++)
590	{
591	if ((pShwPD->a[i].u & (EPT_E_READ \| EPT_E_LEAF)) == (EPT_E_READ \| EPT_E_LEAF))
592	{
593	pShwPD->a[i].u = 0;
594	Assert(pPage->cPresent);
595	pPage->cPresent--;
596	}
597	}
598	if (pPage->cPresent == 0)
599	pPage->iFirstPresent = NIL_PGMPOOL_PRESENT_INDEX;
600	}
601	goto default_case;
602
603	# ifdef VBOX_WITH_NESTED_HWVIRT_VMX_EPT
604	case PGMPOOLKIND_EPT_PD_FOR_EPT_PD: /* Large pages reference 2 MB of physical memory, so we must clear them. */
605	if (pPage->cPresent)
606	{
607	PEPTPD pShwPD = (PEPTPD)PGMPOOL_PAGE_2_PTR_V2(pPool->CTX_SUFF(pVM), pVCpu, pPage);
608	for (unsigned i = 0; i < RT_ELEMENTS(pShwPD->a); i++)
609	{
610	if ( (pShwPD->a[i].u & EPT_PRESENT_MASK)
611	&& (pShwPD->a[i].u & EPT_E_LEAF))
612	{
613	pShwPD->a[i].u = 0;
614	Assert(pPage->cPresent);
615	pPage->cPresent--;
616	}
617	}
618	if (pPage->cPresent == 0)
619	pPage->iFirstPresent = NIL_PGMPOOL_PRESENT_INDEX;
620	}
621	goto default_case;
622	# endif /* VBOX_WITH_NESTED_HWVIRT_VMX_EPT */
623	#endif /* PGM_WITH_LARGE_PAGES */
624
625	case PGMPOOLKIND_32BIT_PT_FOR_32BIT_PT:
626	case PGMPOOLKIND_32BIT_PT_FOR_32BIT_4MB:
627	case PGMPOOLKIND_PAE_PT_FOR_32BIT_PT:
628	case PGMPOOLKIND_PAE_PT_FOR_32BIT_4MB:
629	case PGMPOOLKIND_PAE_PT_FOR_PAE_PT:
630	case PGMPOOLKIND_PAE_PT_FOR_PAE_2MB:
631	case PGMPOOLKIND_32BIT_PT_FOR_PHYS:
632	case PGMPOOLKIND_PAE_PT_FOR_PHYS:
633	case PGMPOOLKIND_EPT_PT_FOR_PHYS:
634	#ifdef VBOX_WITH_NESTED_HWVIRT_VMX_EPT
635	case PGMPOOLKIND_EPT_PT_FOR_EPT_PT:
636	case PGMPOOLKIND_EPT_PT_FOR_EPT_2MB:
637	case PGMPOOLKIND_EPT_PDPT_FOR_EPT_PDPT:
638	case PGMPOOLKIND_EPT_PML4_FOR_EPT_PML4:
639	#endif
640	{
641	if (pPage->cPresent)
642	{
643	void *pvShw = PGMPOOL_PAGE_2_PTR_V2(pPool->CTX_SUFF(pVM), pVCpu, pPage);
644	STAM_PROFILE_START(&pPool->StatZeroPage, z);
645	#if 0
646	/* Useful check for leaking references; very expensive though. */
647	switch (pPage->enmKind)
648	{
649	case PGMPOOLKIND_PAE_PT_FOR_32BIT_PT:
650	case PGMPOOLKIND_PAE_PT_FOR_32BIT_4MB:
651	case PGMPOOLKIND_PAE_PT_FOR_PAE_PT:
652	case PGMPOOLKIND_PAE_PT_FOR_PAE_2MB:
653	case PGMPOOLKIND_PAE_PT_FOR_PHYS:
654	{
655	bool fFoundFirst = false;
656	PPGMSHWPTPAE pPT = (PPGMSHWPTPAE)pvShw;
657	for (unsigned ptIndex = 0; ptIndex < RT_ELEMENTS(pPT->a); ptIndex++)
658	{
659	if (pPT->a[ptIndex].u)
660	{
661	if (!fFoundFirst)
662	{
663	AssertFatalMsg(pPage->iFirstPresent <= ptIndex, ("ptIndex = %d first present = %d\n", ptIndex, pPage->iFirstPresent));
664	if (pPage->iFirstPresent != ptIndex)
665	Log(("ptIndex = %d first present = %d\n", ptIndex, pPage->iFirstPresent));
666	fFoundFirst = true;
667	}
668	if (PGMSHWPTEPAE_IS_P(pPT->a[ptIndex]))
669	{
670	pgmPoolTracDerefGCPhysHint(pPool, pPage, PGMSHWPTEPAE_GET_HCPHYS(pPT->a[ptIndex]), NIL_RTGCPHYS);
671	if (pPage->iFirstPresent == ptIndex)
672	pPage->iFirstPresent = NIL_PGMPOOL_PRESENT_INDEX;
673	}
674	}
675	}
676	AssertFatalMsg(pPage->cPresent == 0, ("cPresent = %d pPage = %RGv\n", pPage->cPresent, pPage->GCPhys));
677	break;
678	}
679	default:
680	break;
681	}
682	#endif
683	ASMMemZeroPage(pvShw);
684	STAM_PROFILE_STOP(&pPool->StatZeroPage, z);
685	pPage->cPresent = 0;
686	pPage->iFirstPresent = NIL_PGMPOOL_PRESENT_INDEX;
687	}
688	}
689	RT_FALL_THRU();
690	default:
691	#ifdef PGM_WITH_LARGE_PAGES
692	default_case:
693	#endif
694	Assert(!pPage->cModifications \|\| ++cModifiedPages);
695	Assert(pPage->iModifiedNext == NIL_PGMPOOL_IDX \|\| pPage->cModifications);
696	Assert(pPage->iModifiedPrev == NIL_PGMPOOL_IDX \|\| pPage->cModifications);
697	pPage->iModifiedNext = NIL_PGMPOOL_IDX;
698	pPage->iModifiedPrev = NIL_PGMPOOL_IDX;
699	pPage->cModifications = 0;
700	break;
701
702	}
703	if (!--cLeft)
704	break;
705	}
706	}
707
708	#ifndef DEBUG_michael
709	AssertMsg(cModifiedPages == pPool->cModifiedPages, ("%d != %d\n", cModifiedPages, pPool->cModifiedPages));
710	#endif
711	pPool->iModifiedHead = NIL_PGMPOOL_IDX;
712	pPool->cModifiedPages = 0;
713
714	/*
715	* Clear all the GCPhys links and rebuild the phys ext free list.
716	*/
717	for (PPGMRAMRANGE pRam = pPool->CTX_SUFF(pVM)->pgm.s.CTX_SUFF(pRamRangesX);
718	pRam;
719	pRam = pRam->CTX_SUFF(pNext))
720	{
721	iPage = pRam->cb >> GUEST_PAGE_SHIFT;
722	while (iPage-- > 0)
723	PGM_PAGE_SET_TRACKING(pVM, &pRam->aPages[iPage], 0);
724	}
725
726	pPool->iPhysExtFreeHead = 0;
727	PPGMPOOLPHYSEXT paPhysExts = pPool->CTX_SUFF(paPhysExts);
728	const unsigned cMaxPhysExts = pPool->cMaxPhysExts;
729	for (unsigned i = 0; i < cMaxPhysExts; i++)
730	{
731	paPhysExts[i].iNext = i + 1;
732	paPhysExts[i].aidx[0] = NIL_PGMPOOL_IDX;
733	paPhysExts[i].apte[0] = NIL_PGMPOOL_PHYSEXT_IDX_PTE;
734	paPhysExts[i].aidx[1] = NIL_PGMPOOL_IDX;
735	paPhysExts[i].apte[1] = NIL_PGMPOOL_PHYSEXT_IDX_PTE;
736	paPhysExts[i].aidx[2] = NIL_PGMPOOL_IDX;
737	paPhysExts[i].apte[2] = NIL_PGMPOOL_PHYSEXT_IDX_PTE;
738	}
739	paPhysExts[cMaxPhysExts - 1].iNext = NIL_PGMPOOL_PHYSEXT_INDEX;
740
741
742	#ifdef PGMPOOL_WITH_OPTIMIZED_DIRTY_PT
743	/* Reset all dirty pages to reactivate the page monitoring. */
744	/* Note: we must do this after clearing all page references and shadow page tables as there might be stale references to
745	* recently removed MMIO ranges around that might otherwise end up asserting in pgmPoolTracDerefGCPhysHint
746	*/
747	for (unsigned i = 0; i < RT_ELEMENTS(pPool->aDirtyPages); i++)
748	{
749	unsigned idxPage = pPool->aidxDirtyPages[i];
750	if (idxPage == NIL_PGMPOOL_IDX)
751	continue;
752
753	PPGMPOOLPAGE pPage = &pPool->aPages[idxPage];
754	Assert(pPage->idx == idxPage);
755	Assert(pPage->iMonitoredNext == NIL_PGMPOOL_IDX && pPage->iMonitoredPrev == NIL_PGMPOOL_IDX);
756
757	AssertMsg(pPage->fDirty, ("Page %RGp (slot=%d) not marked dirty!", pPage->GCPhys, i));
758
759	Log(("Reactivate dirty page %RGp\n", pPage->GCPhys));
760
761	/* First write protect the page again to catch all write accesses. (before checking for changes -> SMP) */
762	int rc = PGMHandlerPhysicalReset(pVM, pPage->GCPhys & ~(RTGCPHYS)GUEST_PAGE_OFFSET_MASK);
763	AssertRCSuccess(rc);
764	pPage->fDirty = false;
765
766	pPool->aidxDirtyPages[i] = NIL_PGMPOOL_IDX;
767	}
768
769	/* Clear all dirty pages. */
770	pPool->idxFreeDirtyPage = 0;
771	pPool->cDirtyPages = 0;
772	#endif
773
774	/* Clear the PGM_SYNC_CLEAR_PGM_POOL flag on all VCPUs to prevent redundant flushes. */
775	for (VMCPUID idCpu = 0; idCpu < pVM->cCpus; idCpu++)
776	pVM->apCpusR3[idCpu]->pgm.s.fSyncFlags &= ~PGM_SYNC_CLEAR_PGM_POOL;
777
778	/* Flush job finished. */
779	VM_FF_CLEAR(pVM, VM_FF_PGM_POOL_FLUSH_PENDING);
780	pPool->cPresent = 0;
781	PGM_UNLOCK(pVM);
782
783	PGM_INVL_ALL_VCPU_TLBS(pVM);
784
785	if (fpvFlushRemTlb)
786	for (VMCPUID idCpu = 0; idCpu < pVM->cCpus; idCpu++)
787	CPUMSetChangedFlags(pVM->apCpusR3[idCpu], CPUM_CHANGED_GLOBAL_TLB_FLUSH);
788
789	STAM_PROFILE_STOP(&pPool->StatClearAll, c);
790	return VINF_SUCCESS;
791	}
792
793
794	/**
795	* Clears the shadow page pool.
796	*
797	* @param pVM The cross context VM structure.
798	* @param fFlushRemTlb When set, the REM TLB is scheduled for flushing as
799	* well.
800	*/
801	void pgmR3PoolClearAll(PVM pVM, bool fFlushRemTlb)
802	{
803	int rc = VMMR3EmtRendezvous(pVM, VMMEMTRENDEZVOUS_FLAGS_TYPE_ONCE, pgmR3PoolClearAllRendezvous, &fFlushRemTlb);
804	AssertRC(rc);
805	}
806
807
808	/**
809	* Stringifies a PGMPOOLACCESS value.
810	*/
811	static const char *pgmPoolPoolAccessToStr(uint8_t enmAccess)
812	{
813	switch ((PGMPOOLACCESS)enmAccess)
814	{
815	case PGMPOOLACCESS_DONTCARE: return "DONTCARE";
816	case PGMPOOLACCESS_USER_RW: return "USER_RW";
817	case PGMPOOLACCESS_USER_R: return "USER_R";
818	case PGMPOOLACCESS_USER_RW_NX: return "USER_RW_NX";
819	case PGMPOOLACCESS_USER_R_NX: return "USER_R_NX";
820	case PGMPOOLACCESS_SUPERVISOR_RW: return "SUPERVISOR_RW";
821	case PGMPOOLACCESS_SUPERVISOR_R: return "SUPERVISOR_R";
822	case PGMPOOLACCESS_SUPERVISOR_RW_NX: return "SUPERVISOR_RW_NX";
823	case PGMPOOLACCESS_SUPERVISOR_R_NX: return "SUPERVISOR_R_NX";
824	}
825	return "Unknown Access";
826	}
827
828
829	/**
830	* Stringifies a PGMPOOLKIND value.
831	*/
832	static const char *pgmPoolPoolKindToStr(uint8_t enmKind)
833	{
834	switch ((PGMPOOLKIND)enmKind)
835	{
836	case PGMPOOLKIND_INVALID:
837	return "INVALID";
838	case PGMPOOLKIND_FREE:
839	return "FREE";
840	case PGMPOOLKIND_32BIT_PT_FOR_PHYS:
841	return "32BIT_PT_FOR_PHYS";
842	case PGMPOOLKIND_32BIT_PT_FOR_32BIT_PT:
843	return "32BIT_PT_FOR_32BIT_PT";
844	case PGMPOOLKIND_32BIT_PT_FOR_32BIT_4MB:
845	return "32BIT_PT_FOR_32BIT_4MB";
846	case PGMPOOLKIND_PAE_PT_FOR_PHYS:
847	return "PAE_PT_FOR_PHYS";
848	case PGMPOOLKIND_PAE_PT_FOR_32BIT_PT:
849	return "PAE_PT_FOR_32BIT_PT";
850	case PGMPOOLKIND_PAE_PT_FOR_32BIT_4MB:
851	return "PAE_PT_FOR_32BIT_4MB";
852	case PGMPOOLKIND_PAE_PT_FOR_PAE_PT:
853	return "PAE_PT_FOR_PAE_PT";
854	case PGMPOOLKIND_PAE_PT_FOR_PAE_2MB:
855	return "PAE_PT_FOR_PAE_2MB";
856	case PGMPOOLKIND_32BIT_PD:
857	return "32BIT_PD";
858	case PGMPOOLKIND_32BIT_PD_PHYS:
859	return "32BIT_PD_PHYS";
860	case PGMPOOLKIND_PAE_PD0_FOR_32BIT_PD:
861	return "PAE_PD0_FOR_32BIT_PD";
862	case PGMPOOLKIND_PAE_PD1_FOR_32BIT_PD:
863	return "PAE_PD1_FOR_32BIT_PD";
864	case PGMPOOLKIND_PAE_PD2_FOR_32BIT_PD:
865	return "PAE_PD2_FOR_32BIT_PD";
866	case PGMPOOLKIND_PAE_PD3_FOR_32BIT_PD:
867	return "PAE_PD3_FOR_32BIT_PD";
868	case PGMPOOLKIND_PAE_PD_FOR_PAE_PD:
869	return "PAE_PD_FOR_PAE_PD";
870	case PGMPOOLKIND_PAE_PD_PHYS:
871	return "PAE_PD_PHYS";
872	case PGMPOOLKIND_PAE_PDPT_FOR_32BIT:
873	return "PAE_PDPT_FOR_32BIT";
874	case PGMPOOLKIND_PAE_PDPT:
875	return "PAE_PDPT";
876	case PGMPOOLKIND_PAE_PDPT_PHYS:
877	return "PAE_PDPT_PHYS";
878	case PGMPOOLKIND_64BIT_PDPT_FOR_64BIT_PDPT:
879	return "64BIT_PDPT_FOR_64BIT_PDPT";
880	case PGMPOOLKIND_64BIT_PDPT_FOR_PHYS:
881	return "64BIT_PDPT_FOR_PHYS";
882	case PGMPOOLKIND_64BIT_PD_FOR_64BIT_PD:
883	return "64BIT_PD_FOR_64BIT_PD";
884	case PGMPOOLKIND_64BIT_PD_FOR_PHYS:
885	return "64BIT_PD_FOR_PHYS";
886	case PGMPOOLKIND_64BIT_PML4:
887	return "64BIT_PML4";
888	case PGMPOOLKIND_EPT_PDPT_FOR_PHYS:
889	return "EPT_PDPT_FOR_PHYS";
890	case PGMPOOLKIND_EPT_PD_FOR_PHYS:
891	return "EPT_PD_FOR_PHYS";
892	case PGMPOOLKIND_EPT_PT_FOR_PHYS:
893	return "EPT_PT_FOR_PHYS";
894	case PGMPOOLKIND_ROOT_NESTED:
895	return "ROOT_NESTED";
896	case PGMPOOLKIND_EPT_PT_FOR_EPT_PT:
897	return "EPT_PT_FOR_EPT_PT";
898	case PGMPOOLKIND_EPT_PT_FOR_EPT_2MB:
899	return "EPT_PT_FOR_EPT_2MB";
900	case PGMPOOLKIND_EPT_PD_FOR_EPT_PD:
901	return "EPT_PD_FOR_EPT_PD";
902	case PGMPOOLKIND_EPT_PDPT_FOR_EPT_PDPT:
903	return "EPT_PDPT_FOR_EPT_PDPT";
904	case PGMPOOLKIND_EPT_PML4_FOR_EPT_PML4:
905	return "EPT_PML4_FOR_EPT_PML4";
906	}
907	return "Unknown kind!";
908	}
909
910
911	/**
912	* Protect all pgm pool page table entries to monitor writes
913	*
914	* @param pVM The cross context VM structure.
915	*
916	* @remarks ASSUMES the caller will flush all TLBs!!
917	*/
918	void pgmR3PoolWriteProtectPages(PVM pVM)
919	{
920	PGM_LOCK_ASSERT_OWNER(pVM);
921	PPGMPOOL pPool = pVM->pgm.s.CTX_SUFF(pPool);
922	unsigned cLeft = pPool->cUsedPages;
923	unsigned iPage = pPool->cCurPages;
924	while (--iPage >= PGMPOOL_IDX_FIRST)
925	{
926	PPGMPOOLPAGE pPage = &pPool->aPages[iPage];
927	if ( pPage->GCPhys != NIL_RTGCPHYS
928	&& pPage->cPresent)
929	{
930	union
931	{
932	void *pv;
933	PX86PT pPT;
934	PPGMSHWPTPAE pPTPae;
935	PEPTPT pPTEpt;
936	} uShw;
937	uShw.pv = PGMPOOL_PAGE_2_PTR(pVM, pPage);
938
939	switch (pPage->enmKind)
940	{
941	/*
942	* We only care about shadow page tables.
943	*/
944	case PGMPOOLKIND_32BIT_PT_FOR_32BIT_PT:
945	case PGMPOOLKIND_32BIT_PT_FOR_32BIT_4MB:
946	case PGMPOOLKIND_32BIT_PT_FOR_PHYS:
947	for (unsigned iShw = 0; iShw < RT_ELEMENTS(uShw.pPT->a); iShw++)
948	if (uShw.pPT->a[iShw].u & X86_PTE_P)
949	uShw.pPT->a[iShw].u = ~(X86PGUINT)X86_PTE_RW;
950	break;
951
952	case PGMPOOLKIND_PAE_PT_FOR_32BIT_PT:
953	case PGMPOOLKIND_PAE_PT_FOR_32BIT_4MB:
954	case PGMPOOLKIND_PAE_PT_FOR_PAE_PT:
955	case PGMPOOLKIND_PAE_PT_FOR_PAE_2MB:
956	case PGMPOOLKIND_PAE_PT_FOR_PHYS:
957	for (unsigned iShw = 0; iShw < RT_ELEMENTS(uShw.pPTPae->a); iShw++)
958	if (PGMSHWPTEPAE_IS_P(uShw.pPTPae->a[iShw]))
959	PGMSHWPTEPAE_SET_RO(uShw.pPTPae->a[iShw]);
960	break;
961
962	case PGMPOOLKIND_EPT_PT_FOR_PHYS:
963	for (unsigned iShw = 0; iShw < RT_ELEMENTS(uShw.pPTEpt->a); iShw++)
964	if (uShw.pPTEpt->a[iShw].u & EPT_E_READ)
965	uShw.pPTEpt->a[iShw].u &= ~(X86PGPAEUINT)EPT_E_WRITE;
966	break;
967
968	default:
969	break;
970	}
971	if (!--cLeft)
972	break;
973	}
974	}
975	}
976
977
978	/**
979	* @callback_method_impl{FNDBGFHANDLERINT, pgmpoolpages}
980	*/
981	static DECLCALLBACK(void) pgmR3PoolInfoPages(PVM pVM, PCDBGFINFOHLP pHlp, const char *pszArgs)
982	{
983	RT_NOREF(pszArgs);
984
985	PPGMPOOL const pPool = pVM->pgm.s.CTX_SUFF(pPool);
986	unsigned const cPages = pPool->cCurPages;
987	unsigned cLeft = pPool->cUsedPages;
988	for (unsigned iPage = 0; iPage < cPages; iPage++)
989	{
990	PGMPOOLPAGE volatile const pPage = (PGMPOOLPAGE volatile const )&pPool->aPages[iPage];
991	RTGCPHYS const GCPhys = pPage->GCPhys;
992	uint8_t const enmKind = pPage->enmKind;
993	if ( enmKind != PGMPOOLKIND_INVALID
994	&& enmKind != PGMPOOLKIND_FREE)
995	{
996	pHlp->pfnPrintf(pHlp, "#%04x: HCPhys=%RHp GCPhys=%RGp %s %s %s%s%s\n",
997	iPage, pPage->Core.Key, GCPhys, pPage->fA20Enabled ? "A20 " : "!A20",
998	pgmPoolPoolKindToStr(enmKind),
999	pPage->enmAccess == PGMPOOLACCESS_DONTCARE ? "" : pgmPoolPoolAccessToStr(pPage->enmAccess),
1000	pPage->fCached ? " cached" : "", pPage->fMonitored ? " monitored" : "");
1001	if (!--cLeft)
1002	break;
1003	}
1004	}
1005	}
1006
1007
1008	/**
1009	* @callback_method_impl{FNDBGFHANDLERINT, pgmpoolroots}
1010	*/
1011	static DECLCALLBACK(void) pgmR3PoolInfoRoots(PVM pVM, PCDBGFINFOHLP pHlp, const char *pszArgs)
1012	{
1013	RT_NOREF(pszArgs);
1014
1015	PPGMPOOL const pPool = pVM->pgm.s.CTX_SUFF(pPool);
1016	unsigned const cPages = pPool->cCurPages;
1017	unsigned cLeft = pPool->cUsedPages;
1018	for (unsigned iPage = 0; iPage < cPages; iPage++)
1019	{
1020	PGMPOOLPAGE volatile const pPage = (PGMPOOLPAGE volatile const )&pPool->aPages[iPage];
1021	RTGCPHYS const GCPhys = pPage->GCPhys;
1022	if (GCPhys != NIL_RTGCPHYS)
1023	{
1024	uint8_t const enmKind = pPage->enmKind;
1025	switch (enmKind)
1026	{
1027	default:
1028	break;
1029
1030	case PGMPOOLKIND_PAE_PDPT_FOR_32BIT:
1031	case PGMPOOLKIND_PAE_PDPT:
1032	case PGMPOOLKIND_PAE_PDPT_PHYS:
1033	case PGMPOOLKIND_64BIT_PML4:
1034	case PGMPOOLKIND_ROOT_NESTED:
1035	case PGMPOOLKIND_EPT_PML4_FOR_EPT_PML4:
1036	pHlp->pfnPrintf(pHlp, "#%04x: HCPhys=%RHp GCPhys=%RGp %s %s %s\n",
1037	iPage, pPage->Core.Key, GCPhys, pPage->fA20Enabled ? "A20 " : "!A20",
1038	pgmPoolPoolKindToStr(enmKind), pPage->fMonitored ? " monitored" : "");
1039	break;
1040	}
1041	if (!--cLeft)
1042	break;
1043	}
1044	}
1045	}
1046
1047	#ifdef VBOX_WITH_DEBUGGER
1048
1049	/**
1050	* Helper for pgmR3PoolCmdCheck that reports an error.
1051	*/
1052	static void pgmR3PoolCheckError(PPGMPOOLCHECKERSTATE pState, const char *pszFormat, ...)
1053	{
1054	if (pState->fFirstMsg)
1055	{
1056	DBGCCmdHlpPrintf(pState->pCmdHlp, "Checking pool page #%i for %RGp %s\n",
1057	pState->pPage->idx, pState->pPage->GCPhys, pgmPoolPoolKindToStr(pState->pPage->enmKind));
1058	pState->fFirstMsg = false;
1059	}
1060
1061	va_list va;
1062	va_start(va, pszFormat);
1063	pState->pCmdHlp->pfnPrintfV(pState->pCmdHlp, NULL, pszFormat, va);
1064	va_end(va);
1065	}
1066
1067
1068	/**
1069	* @callback_method_impl{FNDBGCCMD, The '.pgmpoolcheck' command.}
1070	*/
1071	static DECLCALLBACK(int) pgmR3PoolCmdCheck(PCDBGCCMD pCmd, PDBGCCMDHLP pCmdHlp, PUVM pUVM, PCDBGCVAR paArgs, unsigned cArgs)
1072	{
1073	DBGC_CMDHLP_REQ_UVM_RET(pCmdHlp, pCmd, pUVM);
1074	PVM pVM = pUVM->pVM;
1075	VM_ASSERT_VALID_EXT_RETURN(pVM, VERR_INVALID_VM_HANDLE);
1076	DBGC_CMDHLP_ASSERT_PARSER_RET(pCmdHlp, pCmd, -1, cArgs == 0);
1077	NOREF(paArgs);
1078
1079	PGM_LOCK_VOID(pVM);
1080	PPGMPOOL pPool = pVM->pgm.s.CTX_SUFF(pPool);
1081	PGMPOOLCHECKERSTATE State = { pCmdHlp, pVM, pPool, NULL, true, 0 };
1082	for (unsigned i = 0; i < pPool->cCurPages; i++)
1083	{
1084	PPGMPOOLPAGE pPage = &pPool->aPages[i];
1085	State.pPage = pPage;
1086	State.fFirstMsg = true;
1087
1088	if (pPage->idx != i)
1089	pgmR3PoolCheckError(&State, "Invalid idx value: %#x, expected %#x", pPage->idx, i);
1090
1091	if (pPage->enmKind == PGMPOOLKIND_FREE)
1092	continue;
1093	if (pPage->enmKind > PGMPOOLKIND_LAST \|\| pPage->enmKind <= PGMPOOLKIND_INVALID)
1094	{
1095	if (pPage->enmKind != PGMPOOLKIND_INVALID \|\| pPage->idx != 0)
1096	pgmR3PoolCheckError(&State, "Invalid enmKind value: %#x\n", pPage->enmKind);
1097	continue;
1098	}
1099
1100	void const *pvGuestPage = NULL;
1101	PGMPAGEMAPLOCK LockPage;
1102	if ( pPage->enmKind != PGMPOOLKIND_EPT_PDPT_FOR_PHYS
1103	&& pPage->enmKind != PGMPOOLKIND_EPT_PD_FOR_PHYS
1104	&& pPage->enmKind != PGMPOOLKIND_EPT_PT_FOR_PHYS
1105	&& pPage->enmKind != PGMPOOLKIND_ROOT_NESTED)
1106	{
1107	int rc = PGMPhysGCPhys2CCPtrReadOnly(pVM, pPage->GCPhys, &pvGuestPage, &LockPage);
1108	if (RT_FAILURE(rc))
1109	{
1110	pgmR3PoolCheckError(&State, "PGMPhysGCPhys2CCPtrReadOnly failed for %RGp: %Rrc\n", pPage->GCPhys, rc);
1111	continue;
1112	}
1113	}
1114	# define HCPHYS_TO_POOL_PAGE(a_HCPhys) (PPGMPOOLPAGE)RTAvloHCPhysGet(&pPool->HCPhysTree, (a_HCPhys))
1115
1116	/*
1117	* Check if something obvious is out of sync.
1118	*/
1119	switch (pPage->enmKind)
1120	{
1121	case PGMPOOLKIND_PAE_PT_FOR_PAE_PT:
1122	{
1123	PCPGMSHWPTPAE const pShwPT = (PCPGMSHWPTPAE)PGMPOOL_PAGE_2_PTR(pPool->CTX_SUFF(pVM), pPage);
1124	PCX86PDPAE const pGstPT = (PCX86PDPAE)pvGuestPage;
1125	for (unsigned j = 0; j < RT_ELEMENTS(pShwPT->a); j++)
1126	if (PGMSHWPTEPAE_IS_P(pShwPT->a[j]))
1127	{
1128	RTHCPHYS HCPhys = NIL_RTHCPHYS;
1129	int rc = PGMPhysGCPhys2HCPhys(pPool->CTX_SUFF(pVM), pGstPT->a[j].u & X86_PTE_PAE_PG_MASK, &HCPhys);
1130	if ( rc != VINF_SUCCESS
1131	\|\| PGMSHWPTEPAE_GET_HCPHYS(pShwPT->a[j]) != HCPhys)
1132	pgmR3PoolCheckError(&State, "Mismatch HCPhys: rc=%Rrc idx=%#x guest %RX64 shw=%RX64 vs %RHp\n",
1133	rc, j, pGstPT->a[j].u, PGMSHWPTEPAE_GET_LOG(pShwPT->a[j]), HCPhys);
1134	else if ( PGMSHWPTEPAE_IS_RW(pShwPT->a[j])
1135	&& !(pGstPT->a[j].u & X86_PTE_RW))
1136	pgmR3PoolCheckError(&State, "Mismatch r/w gst/shw: idx=%#x guest %RX64 shw=%RX64 vs %RHp\n",
1137	j, pGstPT->a[j].u, PGMSHWPTEPAE_GET_LOG(pShwPT->a[j]), HCPhys);
1138	}
1139	break;
1140	}
1141
1142	case PGMPOOLKIND_EPT_PT_FOR_EPT_PT:
1143	{
1144	PCEPTPT const pShwPT = (PCEPTPT)PGMPOOL_PAGE_2_PTR(pPool->CTX_SUFF(pVM), pPage);
1145	PCEPTPT const pGstPT = (PCEPTPT)pvGuestPage;
1146	for (unsigned j = 0; j < RT_ELEMENTS(pShwPT->a); j++)
1147	{
1148	uint64_t const uShw = pShwPT->a[j].u;
1149	if (uShw & EPT_PRESENT_MASK)
1150	{
1151	uint64_t const uGst = pGstPT->a[j].u;
1152	RTHCPHYS HCPhys = NIL_RTHCPHYS;
1153	int rc = PGMPhysGCPhys2HCPhys(pPool->CTX_SUFF(pVM), uGst & EPT_E_PG_MASK, &HCPhys);
1154	if ( rc != VINF_SUCCESS
1155	\|\| (uShw & EPT_E_PG_MASK) != HCPhys)
1156	pgmR3PoolCheckError(&State, "Mismatch HCPhys: rc=%Rrc idx=%#x guest %RX64 shw=%RX64 vs %RHp\n",
1157	rc, j, uGst, uShw, HCPhys);
1158	if ( (uShw & (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE))
1159	!= (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE)
1160	&& ( ((uShw & EPT_E_READ) && !(uGst & EPT_E_READ))
1161	\|\| ((uShw & EPT_E_WRITE) && !(uGst & EPT_E_WRITE))
1162	\|\| ((uShw & EPT_E_EXECUTE) && !(uGst & EPT_E_EXECUTE)) ) )
1163	pgmR3PoolCheckError(&State, "Mismatch r/w/x: idx=%#x guest %RX64 shw=%RX64\n", j, uGst, uShw);
1164	}
1165	}
1166	break;
1167	}
1168
1169	case PGMPOOLKIND_EPT_PT_FOR_EPT_2MB:
1170	{
1171	PCEPTPT const pShwPT = (PCEPTPT)PGMPOOL_PAGE_2_PTR(pPool->CTX_SUFF(pVM), pPage);
1172	for (unsigned j = 0; j < RT_ELEMENTS(pShwPT->a); j++)
1173	{
1174	uint64_t const uShw = pShwPT->a[j].u;
1175	if (uShw & EPT_E_LEAF)
1176	pgmR3PoolCheckError(&State, "Leafness-error: idx=%#x shw=%RX64 (2MB)\n", j, uShw);
1177	else if (uShw & EPT_PRESENT_MASK)
1178	{
1179	RTGCPHYS const GCPhysSubPage = pPage->GCPhys \| (j << PAGE_SHIFT);
1180	RTHCPHYS HCPhys = NIL_RTHCPHYS;
1181	int rc = PGMPhysGCPhys2HCPhys(pPool->CTX_SUFF(pVM), GCPhysSubPage, &HCPhys);
1182	if ( rc != VINF_SUCCESS
1183	\|\| (uShw & EPT_E_PG_MASK) != HCPhys)
1184	pgmR3PoolCheckError(&State, "Mismatch HCPhys: rc=%Rrc idx=%#x guest %RX64 shw=%RX64 vs %RHp\n",
1185	rc, j, GCPhysSubPage, uShw, HCPhys);
1186	}
1187	}
1188	break;
1189	}
1190
1191	case PGMPOOLKIND_EPT_PD_FOR_EPT_PD:
1192	{
1193	PCEPTPD const pShwPD = (PCEPTPD)PGMPOOL_PAGE_2_PTR(pPool->CTX_SUFF(pVM), pPage);
1194	PCEPTPD const pGstPD = (PCEPTPD)pvGuestPage;
1195	for (unsigned j = 0; j < RT_ELEMENTS(pShwPD->a); j++)
1196	{
1197	uint64_t const uShw = pShwPD->a[j].u;
1198	if (uShw & EPT_PRESENT_MASK)
1199	{
1200	uint64_t const uGst = pGstPD->a[j].u;
1201	if (uShw & EPT_E_LEAF)
1202	{
1203	if (!(uGst & EPT_E_LEAF))
1204	pgmR3PoolCheckError(&State, "Leafness-mismatch: idx=%#x guest %RX64 shw=%RX64\n", j, uGst, uShw);
1205	else
1206	{
1207	RTHCPHYS HCPhys = NIL_RTHCPHYS;
1208	int rc = PGMPhysGCPhys2HCPhys(pPool->CTX_SUFF(pVM), uGst & EPT_PDE2M_PG_MASK, &HCPhys);
1209	if ( rc != VINF_SUCCESS
1210	\|\| (uShw & EPT_E_PG_MASK) != HCPhys)
1211	pgmR3PoolCheckError(&State, "Mismatch HCPhys: rc=%Rrc idx=%#x guest %RX64 shw=%RX64 vs %RHp (2MB)\n",
1212	rc, j, uGst, uShw, HCPhys);
1213	}
1214	}
1215	else
1216	{
1217	PPGMPOOLPAGE pSubPage = HCPHYS_TO_POOL_PAGE(uShw & EPT_E_PG_MASK);
1218	if (pSubPage)
1219	{
1220	if ( pSubPage->enmKind != PGMPOOLKIND_EPT_PT_FOR_EPT_PT
1221	&& pSubPage->enmKind != PGMPOOLKIND_EPT_PT_FOR_EPT_2MB)
1222	pgmR3PoolCheckError(&State, "Wrong sub-table type: idx=%#x guest %RX64 shw=%RX64: idxSub=%#x %s\n",
1223	j, uGst, uShw, pSubPage->idx, pgmPoolPoolKindToStr(pSubPage->enmKind));
1224	if (pSubPage->fA20Enabled != pPage->fA20Enabled)
1225	pgmR3PoolCheckError(&State, "Wrong sub-table A20: idx=%#x guest %RX64 shw=%RX64: idxSub=%#x A20=%d, expected %d\n",
1226	j, uGst, uShw, pSubPage->idx, pSubPage->fA20Enabled, pPage->fA20Enabled);
1227	if (pSubPage->GCPhys != (uGst & EPT_E_PG_MASK))
1228	pgmR3PoolCheckError(&State, "Wrong sub-table GCPhys: idx=%#x guest %RX64 shw=%RX64: GCPhys=%#RGp idxSub=%#x\n",
1229	j, uGst, uShw, pSubPage->GCPhys, pSubPage->idx);
1230	}
1231	else
1232	pgmR3PoolCheckError(&State, "sub table not found: idx=%#x shw=%RX64\n", j, uShw);
1233	}
1234	if ( (uShw & (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE))
1235	!= (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE)
1236	&& ( ((uShw & EPT_E_READ) && !(uGst & EPT_E_READ))
1237	\|\| ((uShw & EPT_E_WRITE) && !(uGst & EPT_E_WRITE))
1238	\|\| ((uShw & EPT_E_EXECUTE) && !(uGst & EPT_E_EXECUTE)) ) )
1239	pgmR3PoolCheckError(&State, "Mismatch r/w/x: idx=%#x guest %RX64 shw=%RX64\n",
1240	j, uGst, uShw);
1241	}
1242	}
1243	break;
1244	}
1245
1246	case PGMPOOLKIND_EPT_PDPT_FOR_EPT_PDPT:
1247	{
1248	PCEPTPDPT const pShwPDPT = (PCEPTPDPT)PGMPOOL_PAGE_2_PTR(pPool->CTX_SUFF(pVM), pPage);
1249	PCEPTPDPT const pGstPDPT = (PCEPTPDPT)pvGuestPage;
1250	for (unsigned j = 0; j < RT_ELEMENTS(pShwPDPT->a); j++)
1251	{
1252	uint64_t const uShw = pShwPDPT->a[j].u;
1253	if (uShw & EPT_PRESENT_MASK)
1254	{
1255	uint64_t const uGst = pGstPDPT->a[j].u;
1256	if (uShw & EPT_E_LEAF)
1257	pgmR3PoolCheckError(&State, "No 1GiB shadow pages: idx=%#x guest %RX64 shw=%RX64\n", j, uGst, uShw);
1258	else
1259	{
1260	PPGMPOOLPAGE pSubPage = HCPHYS_TO_POOL_PAGE(uShw & EPT_E_PG_MASK);
1261	if (pSubPage)
1262	{
1263	if (pSubPage->enmKind != PGMPOOLKIND_EPT_PD_FOR_EPT_PD)
1264	pgmR3PoolCheckError(&State, "Wrong sub-table type: idx=%#x guest %RX64 shw=%RX64: idxSub=%#x %s\n",
1265	j, uGst, uShw, pSubPage->idx, pgmPoolPoolKindToStr(pSubPage->enmKind));
1266	if (pSubPage->fA20Enabled != pPage->fA20Enabled)
1267	pgmR3PoolCheckError(&State, "Wrong sub-table A20: idx=%#x guest %RX64 shw=%RX64: idxSub=%#x A20=%d, expected %d\n",
1268	j, uGst, uShw, pSubPage->idx, pSubPage->fA20Enabled, pPage->fA20Enabled);
1269	if (pSubPage->GCPhys != (uGst & EPT_E_PG_MASK))
1270	pgmR3PoolCheckError(&State, "Wrong sub-table GCPhys: idx=%#x guest %RX64 shw=%RX64: GCPhys=%#RGp idxSub=%#x\n",
1271	j, uGst, uShw, pSubPage->GCPhys, pSubPage->idx);
1272	}
1273	else
1274	pgmR3PoolCheckError(&State, "sub table not found: idx=%#x shw=%RX64\n", j, uShw);
1275
1276	}
1277	if ( (uShw & (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE))
1278	!= (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE)
1279	&& ( ((uShw & EPT_E_READ) && !(uGst & EPT_E_READ))
1280	\|\| ((uShw & EPT_E_WRITE) && !(uGst & EPT_E_WRITE))
1281	\|\| ((uShw & EPT_E_EXECUTE) && !(uGst & EPT_E_EXECUTE)) ) )
1282	pgmR3PoolCheckError(&State, "Mismatch r/w/x: idx=%#x guest %RX64 shw=%RX64\n",
1283	j, uGst, uShw);
1284	}
1285	}
1286	break;
1287	}
1288
1289	case PGMPOOLKIND_EPT_PML4_FOR_EPT_PML4:
1290	{
1291	PCEPTPML4 const pShwPML4 = (PCEPTPML4)PGMPOOL_PAGE_2_PTR(pPool->CTX_SUFF(pVM), pPage);
1292	PCEPTPML4 const pGstPML4 = (PCEPTPML4)pvGuestPage;
1293	for (unsigned j = 0; j < RT_ELEMENTS(pShwPML4->a); j++)
1294	{
1295	uint64_t const uShw = pShwPML4->a[j].u;
1296	if (uShw & EPT_PRESENT_MASK)
1297	{
1298	uint64_t const uGst = pGstPML4->a[j].u;
1299	if (uShw & EPT_E_LEAF)
1300	pgmR3PoolCheckError(&State, "No 0.5TiB shadow pages: idx=%#x guest %RX64 shw=%RX64\n", j, uGst, uShw);
1301	else
1302	{
1303	PPGMPOOLPAGE pSubPage = HCPHYS_TO_POOL_PAGE(uShw & EPT_E_PG_MASK);
1304	if (pSubPage)
1305	{
1306	if (pSubPage->enmKind != PGMPOOLKIND_EPT_PDPT_FOR_EPT_PDPT)
1307	pgmR3PoolCheckError(&State, "Wrong sub-table type: idx=%#x guest %RX64 shw=%RX64: idxSub=%#x %s\n",
1308	j, uGst, uShw, pSubPage->idx, pgmPoolPoolKindToStr(pSubPage->enmKind));
1309	if (pSubPage->fA20Enabled != pPage->fA20Enabled)
1310	pgmR3PoolCheckError(&State, "Wrong sub-table A20: idx=%#x guest %RX64 shw=%RX64: idxSub=%#x A20=%d, expected %d\n",
1311	j, uGst, uShw, pSubPage->idx, pSubPage->fA20Enabled, pPage->fA20Enabled);
1312	if (pSubPage->GCPhys != (uGst & EPT_E_PG_MASK))
1313	pgmR3PoolCheckError(&State, "Wrong sub-table GCPhys: idx=%#x guest %RX64 shw=%RX64: GCPhys=%#RGp idxSub=%#x\n",
1314	j, uGst, uShw, pSubPage->GCPhys, pSubPage->idx);
1315	}
1316	else
1317	pgmR3PoolCheckError(&State, "sub table not found: idx=%#x shw=%RX64\n", j, uShw);
1318
1319	}
1320	if ( (uShw & (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE))
1321	!= (EPT_E_READ \| EPT_E_WRITE \| EPT_E_EXECUTE)
1322	&& ( ((uShw & EPT_E_READ) && !(uGst & EPT_E_READ))
1323	\|\| ((uShw & EPT_E_WRITE) && !(uGst & EPT_E_WRITE))
1324	\|\| ((uShw & EPT_E_EXECUTE) && !(uGst & EPT_E_EXECUTE)) ) )
1325	pgmR3PoolCheckError(&State, "Mismatch r/w/x: idx=%#x guest %RX64 shw=%RX64\n",
1326	j, uGst, uShw);
1327	}
1328	}
1329	break;
1330	}
1331	}
1332
1333	#undef HCPHYS_TO_POOL_PAGE
1334	if (pvGuestPage)
1335	PGMPhysReleasePageMappingLock(pVM, &LockPage);
1336	}
1337	PGM_UNLOCK(pVM);
1338
1339	if (State.cErrors > 0)
1340	return DBGCCmdHlpFail(pCmdHlp, pCmd, "Found %#x errors", State.cErrors);
1341	DBGCCmdHlpPrintf(pCmdHlp, "no errors found\n");
1342	return VINF_SUCCESS;
1343	}
1344
1345	#endif /* VBOX_WITH_DEBUGGER */

Note: See TracBrowser for help on using the repository browser.

source: vbox/trunk/src/VBox/VMM/VMMR3/PGMPool.cpp@ 97178

Download in other formats: