dosbox-x/include/paging.h
Jonathan Campbell c487bdd002 cleanup
2025-01-23 17:20:00 -08:00

574 lines
20 KiB
C++

/*
* Copyright (C) 2002-2021 The DOSBox Team
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*/
#ifndef DOSBOX_PAGING_H
#define DOSBOX_PAGING_H
#include <exception>
#include "mem.h"
#include "bitop.h"
class PageHandler;
class MEM_CalloutObject;
class PageDirectory;
typedef PageHandler* (MEM_CalloutHandler)(MEM_CalloutObject &co,Bitu phys_page);
void MEM_RegisterHandler(Bitu phys_page,PageHandler *handler,Bitu page_range=1);
void MEM_FreeHandler(Bitu phys_page,Bitu page_range=1);
void MEM_InvalidateCachedHandler(Bitu phys_page,Bitu range=1);
static const Bitu MEMMASK_ISA_20BIT = 0x000000FFU; /* ISA 20-bit decode (20 - 12 = 8) */
static const Bitu MEMMASK_ISA_24BIT = 0x00000FFFU; /* ISA 24-bit decode (24 - 12 = 12) */
static const Bitu MEMMASK_FULL = 0x000FFFFFU; /* full 32-bit decode (32 - 12 = 20) */
extern uint8_t enable_pse_extmask;
/* WARNING: Will only produce a correct result if 'x' is a nonzero power of two.
* For use with MEMMASK_Combine. 'x' is in units of PAGES not BYTES.
*/
static inline Bitu MEMMASK_Range(const Bitu x) {
return ~(x - 1);
}
/* combine range mask with MEMMASK value.
*/
static inline Bitu MEMMASK_Combine(const Bitu a,const Bitu b) {
return a & b;
}
#define MEM_PAGE_SIZE (4096)
#define XMS_START (0x110)
#define TLB_SIZE (1024*1024)
#define PFLAG_READABLE 0x1u
#define PFLAG_WRITEABLE 0x2u
#define PFLAG_HASROM 0x4u
#define PFLAG_HASCODE32 0x8u //Page contains dynamic code
#define PFLAG_NOCODE 0x10u //No dynamic code can be generated here
#define PFLAG_INIT 0x20u //No dynamic code can be generated here
#define PFLAG_HASCODE16 0x40u //Page contains 16-bit dynamic code
#define PFLAG_HASCODE (PFLAG_HASCODE32|PFLAG_HASCODE16)
#define LINK_START ((1024+64)/4) //Start right after the HMA
//Allow 128 mb of memory to be linked
#define PAGING_LINKS (128*1024/4)
/* NTS: Despite that all devices respond to these page handlers, all addresses are still linear addresses.
* Devices that need physical memory addresses will need to call PAGING_GetPhysicalAddress() to translate
* linear to physical, or PAGING_GetPhysicalAddress64() if the device supports responding to addresses
* above 4GB. This sounds crazy, but this is one aspect of emulation inherited from DOSBox SVN and I
* don't think it's wise to break code to change it. */
class PageHandler {
public:
PageHandler(Bitu flg) : flags(flg) {}
virtual ~PageHandler(void) { }
PageHandler(void) : flags(0) { }
virtual uint8_t readb(PhysPt addr);
virtual uint16_t readw(PhysPt addr);
virtual uint32_t readd(PhysPt addr);
virtual void writeb(PhysPt addr,uint8_t val);
virtual void writew(PhysPt addr,uint16_t val);
virtual void writed(PhysPt addr,uint32_t val);
virtual HostPt GetHostReadPt(PageNum phys_page);
virtual HostPt GetHostWritePt(PageNum phys_page);
virtual bool readb_checked(PhysPt addr,uint8_t * val);
virtual bool readw_checked(PhysPt addr,uint16_t * val);
virtual bool readd_checked(PhysPt addr,uint32_t * val);
virtual bool writeb_checked(PhysPt addr,uint8_t val);
virtual bool writew_checked(PhysPt addr,uint16_t val);
virtual bool writed_checked(PhysPt addr,uint32_t val);
Bitu getFlags() const {
return flags;
}
void setFlags(const Bitu flagsNew) {
flags = flagsNew;
}
Bitu flags;
private:
PageHandler(const PageHandler&);
};
/* NTS: To explain the Install() method, the caller not only provides the IOMASK_.. value, but ANDs
* the least significant bits to define the range of I/O ports to respond to. An ISA Sound Blaster
* for example would set portmask = (IOMASK_ISA_10BIT & (~0xF)) in order to respond to 220h-22Fh,
* 240h-24Fh, etc. At I/O callout time, the callout object is tested
* if (cpu_ioport & io_mask) == (m_port & io_mask)
*
* This does not prevent emulation of devices that start on non-aligned ports or strange port ranges,
* because the callout handler is free to decline the I/O request, leading the callout process to
* move on to the next device or mark the I/O port as empty. */
class MEM_CalloutObject {
public:
MEM_CalloutObject() {};
void InvalidateCachedHandlers(void);
void Install(Bitu page,Bitu pagemask/*MEMMASK_ISA_24BIT, etc.*/,MEM_CalloutHandler *handler);
void Uninstall();
public:
bool installed = false;
Bitu mem_mask = 0xFFFFFFFF;
Bitu range_mask = 0;
Bitu alias_mask = 0xFFFFFFFF;
unsigned int getcounter = 0;
MEM_CalloutHandler *m_handler = NULL;
Bitu m_base = 0;
bool alloc = false;
public:
inline bool MatchPage(const Bitu p) {
/* (p & io_mask) == (m_port & io_mask) but this also works.
* apparently modern x86 processors are faster at addition/subtraction than bitmasking.
* for this to work, m_port must be a multiple of the I/O range. For example, if the I/O
* range is 16 ports, then m_port must be a multiple of 16. */
return ((p - m_base) & mem_mask) == 0;
}
inline bool isInstalled(void) {
return installed;
}
};
enum MEM_Type_t {
MEM_TYPE_NONE=0,
MEM_TYPE_MIN=1,
MEM_TYPE_ISA=1,
MEM_TYPE_PCI,
MEM_TYPE_MB,
MEM_TYPE_MAX
};
void MEM_InitCallouts(void);
typedef uint32_t MEM_Callout_t;
static inline uint32_t MEM_Callout_t_comb(const enum MEM_Type_t t,const uint32_t idx) {
return ((uint32_t)t << (uint32_t)28) + idx;
}
static inline enum MEM_Type_t MEM_Callout_t_type(const MEM_Callout_t t) {
return (enum MEM_Type_t)(t >> 28);
}
static inline uint32_t MEM_Callout_t_index(const MEM_Callout_t t) {
return t & (((uint32_t)1 << (uint32_t)28) - (uint32_t)1);
}
static const MEM_Callout_t MEM_Callout_t_none = (MEM_Callout_t)0;
MEM_Callout_t MEM_AllocateCallout(MEM_Type_t t);
void MEM_FreeCallout(MEM_Callout_t c);
MEM_CalloutObject *MEM_GetCallout(MEM_Callout_t c);
void MEM_PutCallout(MEM_CalloutObject *obj);
/* Some other functions */
void PAGING_Enable(bool enabled);
bool PAGING_Enabled(void);
void PAGING_SetWP(bool wp);
void PAGING_SwitchCPL(bool isUser);
Bitu PAGING_GetDirBase(void);
void PAGING_SetDirBase(Bitu cr3);
void PAGING_InitTLB(void);
void PAGING_ClearTLB(void);
void PAGING_LinkPage(PageNum lin_page,PageNum phys_page);
void PAGING_UnlinkPages(PageNum lin_page,PageNum pages);
/* This maps the page directly, only use when paging is disabled */
void PAGING_MapPage(PageNum lin_page,PageNum phys_page);
bool PAGING_MakePhysPage(PageNum &page);
void MEM_SetLFB(Bitu page, Bitu pages, PageHandler *handler, PageHandler *mmiohandler);
void MEM_SetPageHandler(Bitu phys_page, Bitu pages, PageHandler * handler);
#ifdef _MSC_VER
#pragma pack (1)
#endif
/* Page table TLB. At some point, 64-bit builds will have 64-bit TLB type */
typedef uint32_t tlbentry_t;
static constexpr size_t tlbentry_bits = bitop::type_bits<tlbentry_t>();
/* bits 31-30: ACCESS_* contants */
static constexpr tlbentry_t PHYSPAGE_ACCESS_BITS_SHIFT = tlbentry_bits - 2u;
static constexpr tlbentry_t PHYSPAGE_ACCESS_BITS = bitop::bitcount2masklsb<2u,PHYSPAGE_ACCESS_BITS_SHIFT,tlbentry_t>();
/* bit 29: dirty bit */
static constexpr tlbentry_t PHYSPAGE_DIRTY_SHIFT = tlbentry_bits - 3u;
static constexpr tlbentry_t PHYSPAGE_DIRTY = bitop::bit2mask<PHYSPAGE_DIRTY_SHIFT,tlbentry_t>();
/* bits 28: not defined */
/* bits 27-0: physical page. 40-bit addresses, that's as far as we can go, and the limits of PSE anyway */
static constexpr tlbentry_t PHYSPAGE_ADDR =
build_memlimit_32bit() ?
bitop::bitcount2masklsb<20,0,tlbentry_t>() : /* 32-bit builds limited to 1GB no point emulating past 4GB (20+12) = 32 */
bitop::bitcount2masklsb<28,0,tlbentry_t>(); /* 64-bit builds allowed up to 1TB (28+12) = 40 */
struct X86_PageEntryBlock{ // Page Table Entry, though it keeps the PageEntryBlock name to avoid breaking all this code
#ifdef WORDS_BIGENDIAN
uint32_t base:20; // [31:12]
uint32_t avl:3; // [11: 8]
uint32_t g:1; // [ 8: 8]
uint32_t pat:1; // [ 7: 7]
uint32_t d:1; // [ 6: 6]
uint32_t a:1; // [ 5: 5]
uint32_t pcd:1; // [ 4: 4]
uint32_t pwt:1; // [ 3: 3]
uint32_t us:1; // [ 2: 2]
uint32_t wr:1; // [ 1: 1]
uint32_t p:1; // [ 0: 0]
#else
uint32_t p:1; // [ 0: 0]
uint32_t wr:1; // [ 1: 1] R/W
uint32_t us:1; // [ 2: 2] U/S
uint32_t pwt:1; // [ 3: 3]
uint32_t pcd:1; // [ 4: 4]
uint32_t a:1; // [ 5: 5]
uint32_t d:1; // [ 6: 6]
uint32_t pat:1; // [ 7: 7]
uint32_t g:1; // [ 8: 8]
uint32_t avl:3; // [11: 9]
uint32_t base:20; // [31:12]
#endif
} GCC_ATTRIBUTE(packed);
struct X86_PageDirEntryBlock{
#ifdef WORDS_BIGENDIAN
uint32_t base:20; // [31:12]
uint32_t avl:3; // [11: 9]
uint32_t g:1; // [ 8: 8]
uint32_t ps:1; // [ 7: 7]
uint32_t avl6:1; // [ 6: 6]
uint32_t a:1; // [ 5: 5]
uint32_t pcd:1; // [ 4: 4]
uint32_t pwt:1; // [ 3: 3]
uint32_t us:1; // [ 2: 2]
uint32_t wr:1; // [ 1: 1]
uint32_t p:1; // [ 0: 0]
#else
uint32_t p:1; // [ 0: 0]
uint32_t wr:1; // [ 1: 1] R/W
uint32_t us:1; // [ 2: 2] U/S
uint32_t pwt:1; // [ 3: 3]
uint32_t pcd:1; // [ 4: 4]
uint32_t a:1; // [ 5: 5]
uint32_t avl6:1; // [ 6: 6] See Note [*1], this bit changed meaning between the 486 and Pentium!
uint32_t ps:1; // [ 7: 7] Page Size Extension (Pentium), even though Intel didn't document it until later
uint32_t g:1; // [ 8: 8]
uint32_t avl:3; // [11: 9]
uint32_t base:20; // [31:12]
#endif
} GCC_ATTRIBUTE(packed);
/* Note [*1]: The i386 defined just the Page Table Entry to represent the PDE and PTE, because they had the same layout.
*
* The i486 lists the PDE and PTE, even though they have the same bit layout.
*
* The Pentium however, seems to have decided that bit 6 of the PDE is no longer the "dirty" bit and is redefined as AVL.
* Although, if PSE is enabled and the PDE is the 4MB format, bit 6 is once again a "dirty" but.
*
* Ref: [http://hackipedia.org/browse.cgi/Computer/Platform/PC%2c%20IBM%20compatible/CPU/80386/Intel/386DX%20Microprocessor%20Programmer%27s%20Reference%20Manual%20%281990%29%2epdf]
* Ref: [http://hackipedia.org/browse.cgi/Computer/Platform/PC%2c%20IBM%20compatible/CPU/80486/Intel/i486%20Microprocessor%20%281989%2d04%29%2epdf]
* Ref: [http://hackipedia.org/browse.cgi/Computer/Platform/PC%2c%20IBM%20compatible/CPU/Pentium/Pentium%20Processor%20Family%20Developer%27s%20Manual%20%2d%20Volume%203%3a%20Architecture%20and%20Programming%20Manual%20%281995%2d07%29%2epdf] */
struct X86_PageDir4MBEntryBlock{ // PSE=1 and PageDirEntryBlock PS=1
#ifdef WORDS_BIGENDIAN
uint32_t base22:10; // [31:22] bits 31:22
uint32_t reserved:1; // [21:21]
uint32_t base32:8; // [20:13] bits 39:32 if PSE36
uint32_t pat:1; // [12:12]
uint32_t avl:3; // [11: 9]
uint32_t g:1; // [ 8: 8]
uint32_t ps:1; // [ 7: 7] PS=1, or else this is just X86_PageDirEntryBlock
uint32_t d:1; // [ 6: 6]
uint32_t a:1; // [ 5: 5]
uint32_t pcd:1; // [ 4: 4]
uint32_t pwt:1; // [ 3: 3]
uint32_t us:1; // [ 2: 2] U/S
uint32_t wr:1; // [ 1: 1] R/W
uint32_t p:1; // [ 0: 0]
#else
uint32_t p:1; // [ 0: 0]
uint32_t wr:1; // [ 1: 1] R/W
uint32_t us:1; // [ 2: 2] U/S
uint32_t pwt:1; // [ 3: 3]
uint32_t pcd:1; // [ 4: 4]
uint32_t a:1; // [ 5: 5]
uint32_t d:1; // [ 6: 6]
uint32_t ps:1; // [ 7: 7] PS=1, or else this is just X86_PageDirEntryBlock
uint32_t g:1; // [ 8: 8]
uint32_t avl:3; // [11: 9]
uint32_t pat:1; // [12:12]
uint32_t base32:8; // [20:13] bits 39:32 if PSE36
uint32_t reserved:1; // [21:21]
uint32_t base22:10; // [31:22] bits 31:22
#endif
inline PageNum getBase(const LinearPt lin_page) const {
return PageNum(base22 << PageNum(10u)) +
PageNum((base32 & enable_pse_extmask) << PageNum(20u)) +
PageNum(lin_page & 0x3FFu);
}
} GCC_ATTRIBUTE(packed);
#ifdef _MSC_VER
#pragma pack ()
#endif
union X86PageEntry {
uint32_t load;
X86_PageEntryBlock block;
X86_PageDirEntryBlock dirblock;
X86_PageDir4MBEntryBlock dirblock4mb; // PSE=1 and PS=1
template <const uint8_t bitpos/*for constant compile time optimization*/> inline uint8_t accbits(void) const {
if (bitpos == 0)
return (load >> 1u) & 3u; /* ex: (load >> 1u) & 0x3 -> U/S:R/W->[1:0]*/
else /* usually 2 */
return (load << (bitpos - 1u)) & (3u << bitpos); /* ex: (load << 1u) & 0xc -> U/S:R/W->[3:2]*/
}
};
static_assert( sizeof(X86PageEntry) == 4, "oops" );
struct PagingBlock {
uint32_t cr3;
uint32_t cr2;
struct {
PageNum page;
PhysPt addr;
} base;
struct {
HostPt read[TLB_SIZE];
HostPt write[TLB_SIZE];
PageHandler *readhandler[TLB_SIZE];
PageHandler *writehandler[TLB_SIZE];
tlbentry_t phys_page[TLB_SIZE];
} tlb;
struct {
uint32_t used;
uint32_t entries[PAGING_LINKS]; /* does not require more than 32 bits */
} links;
struct {
uint32_t used;
uint32_t entries[PAGING_LINKS]; /* does not require more than 32 bits */
} ur_links;
struct {
uint32_t used;
uint32_t entries[PAGING_LINKS]; /* does not require more than 32 bits */
} krw_links;
struct {
uint32_t used;
uint32_t entries[PAGING_LINKS]; /* does not require more than 32 bits */
} kr_links; // WP-only
uint32_t firstmb[LINK_START]; /* does not use flags and does not reach beyond 1MB, does not need more than 32 bits */
bool enabled;
bool wp;
};
extern PagingBlock paging;
/* Some support functions */
PageHandler * MEM_GetPageHandler(const Bitu phys_page);
/* Unaligned address handlers */
uint16_t mem_unalignedreadw(const LinearPt address);
uint32_t mem_unalignedreadd(const LinearPt address);
void mem_unalignedwritew(const LinearPt address,const uint16_t val);
void mem_unalignedwrited(const LinearPt address,const uint32_t val);
bool mem_unalignedreadw_checked(const LinearPt address,uint16_t * const val);
bool mem_unalignedreadd_checked(const LinearPt address,uint32_t * const val);
bool mem_unalignedwritew_checked(const LinearPt address,uint16_t const val);
bool mem_unalignedwrited_checked(const LinearPt address,uint32_t const val);
static INLINE HostPt get_tlb_read(const LinearPt address) {
return paging.tlb.read[address>>12];
}
static INLINE HostPt get_tlb_write(const LinearPt address) {
return paging.tlb.write[address>>12];
}
static INLINE PageHandler* get_tlb_readhandler(const LinearPt address) {
return paging.tlb.readhandler[address>>12];
}
static INLINE PageHandler* get_tlb_writehandler(const LinearPt address) {
return paging.tlb.writehandler[address>>12];
}
/* Use these helper functions to access linear addresses in readX/writeX functions */
/* NTS: 12-bit shift 32-bit constant, upper bits get shifted out, therefore no need to bitmask */
static INLINE PhysPt PAGING_GetPhysicalPage(const LinearPt linePage) {
return (paging.tlb.phys_page[linePage>>12]<<12);
}
static INLINE PhysPt PAGING_GetPhysicalPageNumber(const LinearPt linePage) {
return paging.tlb.phys_page[linePage>>12]&PHYSPAGE_ADDR;
}
/* NTS: 12-bit shift 32-bit constant, upper bits get shifted out, therefore no need to bitmask */
static INLINE PhysPt PAGING_GetPhysicalAddress(const LinearPt linAddr) {
return (paging.tlb.phys_page[linAddr>>12]<<12)|(linAddr&0xfff);
}
static INLINE PhysPt64 PAGING_GetPhysicalAddress64(const LinearPt linAddr) {
return ((PhysPt64)(paging.tlb.phys_page[linAddr>>12]&PHYSPAGE_ADDR)<<(PhysPt64)12)|(linAddr&0xfff);
}
/* Special inlined memory reading/writing */
static INLINE uint8_t mem_readb_inline(const LinearPt address) {
const HostPt tlb_addr=get_tlb_read(address);
if (tlb_addr) return host_readb(tlb_addr+address);
else return (uint8_t)(get_tlb_readhandler(address))->readb(address);
}
static INLINE uint16_t mem_readw_inline(const LinearPt address) {
if ((address & 0xfff)<0xfff) {
const HostPt tlb_addr=get_tlb_read(address);
if (tlb_addr) return host_readw(tlb_addr+address);
else return (uint16_t)(get_tlb_readhandler(address))->readw(address);
} else return mem_unalignedreadw(address);
}
static INLINE uint32_t mem_readd_inline(const LinearPt address) {
if ((address & 0xfff)<0xffd) {
const HostPt tlb_addr=get_tlb_read(address);
if (tlb_addr) return host_readd(tlb_addr+address);
else return (uint32_t)(get_tlb_readhandler(address))->readd(address);
} else return mem_unalignedreadd(address);
}
static INLINE void mem_writeb_inline(const LinearPt address,const uint8_t val) {
const HostPt tlb_addr=get_tlb_write(address);
if (tlb_addr) host_writeb(tlb_addr+address,val);
else (get_tlb_writehandler(address))->writeb(address,val);
}
static INLINE void mem_writew_inline(const LinearPt address,const uint16_t val) {
if ((address & 0xfffu)<0xfffu) {
const HostPt tlb_addr=get_tlb_write(address);
if (tlb_addr) host_writew(tlb_addr+address,val);
else (get_tlb_writehandler(address))->writew(address,val);
} else mem_unalignedwritew(address,val);
}
static INLINE void mem_writed_inline(const LinearPt address,const uint32_t val) {
if ((address & 0xfffu)<0xffdu) {
const HostPt tlb_addr=get_tlb_write(address);
if (tlb_addr) host_writed(tlb_addr+address,val);
else (get_tlb_writehandler(address))->writed(address,val);
} else mem_unalignedwrited(address,val);
}
static INLINE bool mem_readb_checked(const LinearPt address, uint8_t * const val) {
const HostPt tlb_addr=get_tlb_read(address);
if (tlb_addr) {
*val=host_readb(tlb_addr+address);
return false;
} else return (get_tlb_readhandler(address))->readb_checked(address, val);
}
static INLINE bool mem_readw_checked(const LinearPt address, uint16_t * const val) {
if ((address & 0xfffu)<0xfffu) {
const HostPt tlb_addr=get_tlb_read(address);
if (tlb_addr) {
*val=host_readw(tlb_addr+address);
return false;
} else return (get_tlb_readhandler(address))->readw_checked(address, val);
} else return mem_unalignedreadw_checked(address, val);
}
static INLINE bool mem_readd_checked(const LinearPt address, uint32_t * const val) {
if ((address & 0xfffu)<0xffdu) {
const HostPt tlb_addr=get_tlb_read(address);
if (tlb_addr) {
*val=host_readd(tlb_addr+address);
return false;
} else return (get_tlb_readhandler(address))->readd_checked(address, val);
} else return mem_unalignedreadd_checked(address, val);
}
static INLINE bool mem_writeb_checked(const LinearPt address,const uint8_t val) {
const HostPt tlb_addr=get_tlb_write(address);
if (tlb_addr) {
host_writeb(tlb_addr+address,val);
return false;
} else return (get_tlb_writehandler(address))->writeb_checked(address,val);
}
static INLINE bool mem_writew_checked(const LinearPt address,const uint16_t val) {
if ((address & 0xfffu)<0xfffu) {
const HostPt tlb_addr=get_tlb_write(address);
if (tlb_addr) {
host_writew(tlb_addr+address,val);
return false;
} else return (get_tlb_writehandler(address))->writew_checked(address,val);
} else return mem_unalignedwritew_checked(address,val);
}
static INLINE bool mem_writed_checked(const LinearPt address,const uint32_t val) {
if ((address & 0xfffu)<0xffdu) {
const HostPt tlb_addr=get_tlb_write(address);
if (tlb_addr) {
host_writed(tlb_addr+address,val);
return false;
} else return (get_tlb_writehandler(address))->writed_checked(address,val);
} else return mem_unalignedwrited_checked(address,val);
}
extern bool dosbox_allow_nonrecursive_page_fault; /* when set, do nonrecursive mode (when executing instruction) */
class GuestPageFaultException : public std::exception {
public:
const char *what() const throw() override {
return "Guest page fault exception";
}
GuestPageFaultException(LinearPt n_lin_addr, Bitu n_page_addr, Bitu n_faultcode) : lin_addr(n_lin_addr), page_addr(n_page_addr), faultcode(n_faultcode) {
}
public:
LinearPt lin_addr;
Bitu page_addr;
Bitu faultcode;
};
class GuestGenFaultException : public std::exception {
public:
virtual const char *what() const throw() override {
return "Guest general protection fault exception";
}
GuestGenFaultException() {
}
};
uint8_t PageHandler_HostPtReadB(PageHandler *p,PhysPt addr);
uint16_t PageHandler_HostPtReadW(PageHandler *p,PhysPt addr);
uint32_t PageHandler_HostPtReadD(PageHandler *p,PhysPt addr);
void PageHandler_HostPtWriteB(PageHandler *p,PhysPt addr,uint8_t val);
void PageHandler_HostPtWriteW(PageHandler *p,PhysPt addr,uint16_t val);
void PageHandler_HostPtWriteD(PageHandler *p,PhysPt addr,uint32_t val);
#endif