gem5/src/arch/x86/tlb.cc

451 lines
15 KiB
C++
Raw Normal View History

/*
* Copyright (c) 2007-2008 The Hewlett-Packard Development Company
* All rights reserved.
*
* The license below extends only to copyright in the software and shall
* not be construed as granting a license to any other intellectual
* property including but not limited to intellectual property relating
* to a hardware implementation of the functionality of the software
* licensed hereunder. You may use the software subject to the license
* terms below provided that you ensure that this notice is replicated
* unmodified and in its entirety in all distributions of the software,
* modified or unmodified, in source code or in binary form.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met: redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer;
* redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution;
* neither the name of the copyright holders nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* Authors: Gabe Black
*/
#include <cstring>
#include "arch/x86/insts/microldstop.hh"
#include "arch/x86/regs/misc.hh"
#include "arch/x86/regs/msr.hh"
2011-04-15 19:44:06 +02:00
#include "arch/x86/faults.hh"
#include "arch/x86/pagetable.hh"
#include "arch/x86/tlb.hh"
#include "arch/x86/x86_traits.hh"
#include "base/bitfield.hh"
#include "base/trace.hh"
#include "config/full_system.hh"
#include "cpu/base.hh"
2011-04-15 19:44:06 +02:00
#include "cpu/thread_context.hh"
#include "debug/TLB.hh"
#include "mem/packet_access.hh"
#include "mem/request.hh"
#if FULL_SYSTEM
#include "arch/x86/pagetable_walker.hh"
#else
#include "mem/page_table.hh"
#include "sim/process.hh"
#endif
namespace X86ISA {
TLB::TLB(const Params *p) : BaseTLB(p), configAddress(0), size(p->size)
{
tlb = new TlbEntry[size];
std::memset(tlb, 0, sizeof(TlbEntry) * size);
for (int x = 0; x < size; x++)
freeList.push_back(&tlb[x]);
#if FULL_SYSTEM
walker = p->walker;
walker->setTLB(this);
#endif
}
TlbEntry *
TLB::insert(Addr vpn, TlbEntry &entry)
{
//TODO Deal with conflicting entries
TlbEntry *newEntry = NULL;
if (!freeList.empty()) {
newEntry = freeList.front();
freeList.pop_front();
} else {
newEntry = entryList.back();
entryList.pop_back();
}
*newEntry = entry;
newEntry->vaddr = vpn;
entryList.push_front(newEntry);
return newEntry;
}
TLB::EntryList::iterator
TLB::lookupIt(Addr va, bool update_lru)
{
//TODO make this smarter at some point
EntryList::iterator entry;
for (entry = entryList.begin(); entry != entryList.end(); entry++) {
if ((*entry)->vaddr <= va && (*entry)->vaddr + (*entry)->size > va) {
DPRINTF(TLB, "Matched vaddr %#x to entry starting at %#x "
"with size %#x.\n", va, (*entry)->vaddr, (*entry)->size);
if (update_lru) {
entryList.push_front(*entry);
entryList.erase(entry);
entry = entryList.begin();
}
break;
}
}
return entry;
}
TlbEntry *
TLB::lookup(Addr va, bool update_lru)
{
EntryList::iterator entry = lookupIt(va, update_lru);
if (entry == entryList.end())
return NULL;
else
return *entry;
}
void
TLB::invalidateAll()
{
DPRINTF(TLB, "Invalidating all entries.\n");
while (!entryList.empty()) {
TlbEntry *entry = entryList.front();
entryList.pop_front();
freeList.push_back(entry);
}
}
void
TLB::setConfigAddress(uint32_t addr)
{
configAddress = addr;
}
void
TLB::invalidateNonGlobal()
{
DPRINTF(TLB, "Invalidating all non global entries.\n");
EntryList::iterator entryIt;
for (entryIt = entryList.begin(); entryIt != entryList.end();) {
if (!(*entryIt)->global) {
freeList.push_back(*entryIt);
entryList.erase(entryIt++);
} else {
entryIt++;
}
}
}
void
TLB::demapPage(Addr va, uint64_t asn)
{
EntryList::iterator entry = lookupIt(va, false);
if (entry != entryList.end()) {
freeList.push_back(*entry);
entryList.erase(entry);
}
}
Fault
TLB::translateInt(RequestPtr req, ThreadContext *tc)
{
DPRINTF(TLB, "Addresses references internal memory.\n");
Addr vaddr = req->getVaddr();
Addr prefix = (vaddr >> 3) & IntAddrPrefixMask;
if (prefix == IntAddrPrefixCPUID) {
panic("CPUID memory space not yet implemented!\n");
} else if (prefix == IntAddrPrefixMSR) {
vaddr = (vaddr >> 3) & ~IntAddrPrefixMask;
req->setFlags(Request::MMAPPED_IPR);
MiscRegIndex regNum;
if (!msrAddrToIndex(regNum, vaddr))
return new GeneralProtection(0);
//The index is multiplied by the size of a MiscReg so that
//any memory dependence calculations will not see these as
//overlapping.
req->setPaddr((Addr)regNum * sizeof(MiscReg));
return NoFault;
} else if (prefix == IntAddrPrefixIO) {
// TODO If CPL > IOPL or in virtual mode, check the I/O permission
// bitmap in the TSS.
Addr IOPort = vaddr & ~IntAddrPrefixMask;
// Make sure the address fits in the expected 16 bit IO address
// space.
assert(!(IOPort & ~0xFFFF));
if (IOPort == 0xCF8 && req->getSize() == 4) {
req->setFlags(Request::MMAPPED_IPR);
req->setPaddr(MISCREG_PCI_CONFIG_ADDRESS * sizeof(MiscReg));
} else if ((IOPort & ~mask(2)) == 0xCFC) {
req->setFlags(Request::UNCACHEABLE);
Addr configAddress =
tc->readMiscRegNoEffect(MISCREG_PCI_CONFIG_ADDRESS);
if (bits(configAddress, 31, 31)) {
req->setPaddr(PhysAddrPrefixPciConfig |
mbits(configAddress, 30, 2) |
(IOPort & mask(2)));
} else {
req->setPaddr(PhysAddrPrefixIO | IOPort);
}
} else {
req->setFlags(Request::UNCACHEABLE);
req->setPaddr(PhysAddrPrefixIO | IOPort);
}
return NoFault;
} else {
panic("Access to unrecognized internal address space %#x.\n",
prefix);
}
}
Fault
TLB::translate(RequestPtr req, ThreadContext *tc, Translation *translation,
Mode mode, bool &delayedResponse, bool timing)
{
uint32_t flags = req->getFlags();
int seg = flags & SegmentFlagMask;
bool storeCheck = flags & (StoreCheck << FlagShift);
delayedResponse = false;
// If this is true, we're dealing with a request to a non-memory address
// space.
if (seg == SEGMENT_REG_MS) {
return translateInt(req, tc);
}
Addr vaddr = req->getVaddr();
DPRINTF(TLB, "Translating vaddr %#x.\n", vaddr);
HandyM5Reg m5Reg = tc->readMiscRegNoEffect(MISCREG_M5_REG);
// If protected mode has been enabled...
if (m5Reg.prot) {
DPRINTF(TLB, "In protected mode.\n");
// If we're not in 64-bit mode, do protection/limit checks
if (m5Reg.mode != LongMode) {
DPRINTF(TLB, "Not in long mode. Checking segment protection.\n");
// Check for a NULL segment selector.
if (!(seg == SEGMENT_REG_TSG || seg == SYS_SEGMENT_REG_IDTR ||
seg == SEGMENT_REG_HS || seg == SEGMENT_REG_LS)
&& !tc->readMiscRegNoEffect(MISCREG_SEG_SEL(seg)))
return new GeneralProtection(0);
bool expandDown = false;
2009-02-27 18:23:50 +01:00
SegAttr attr = tc->readMiscRegNoEffect(MISCREG_SEG_ATTR(seg));
if (seg >= SEGMENT_REG_ES && seg <= SEGMENT_REG_HS) {
if (!attr.writable && (mode == Write || storeCheck))
return new GeneralProtection(0);
2009-04-09 07:21:27 +02:00
if (!attr.readable && mode == Read)
return new GeneralProtection(0);
expandDown = attr.expandDown;
2009-02-27 18:23:50 +01:00
}
Addr base = tc->readMiscRegNoEffect(MISCREG_SEG_BASE(seg));
Addr limit = tc->readMiscRegNoEffect(MISCREG_SEG_LIMIT(seg));
2009-02-27 18:23:50 +01:00
// This assumes we're not in 64 bit mode. If we were, the default
// address size is 64 bits, overridable to 32.
int size = 32;
bool sizeOverride = (flags & (AddrSizeFlagBit << FlagShift));
SegAttr csAttr = tc->readMiscRegNoEffect(MISCREG_CS_ATTR);
2009-02-28 02:29:58 +01:00
if ((csAttr.defaultSize && sizeOverride) ||
(!csAttr.defaultSize && !sizeOverride))
2009-02-27 18:23:50 +01:00
size = 16;
Addr offset = bits(vaddr - base, size-1, 0);
Addr endOffset = offset + req->getSize() - 1;
if (expandDown) {
DPRINTF(TLB, "Checking an expand down segment.\n");
2009-02-27 18:23:50 +01:00
warn_once("Expand down segments are untested.\n");
if (offset <= limit || endOffset <= limit)
return new GeneralProtection(0);
} else {
2009-02-27 18:23:50 +01:00
if (offset > limit || endOffset > limit)
return new GeneralProtection(0);
}
}
// If paging is enabled, do the translation.
if (m5Reg.paging) {
DPRINTF(TLB, "Paging enabled.\n");
// The vaddr already has the segment base applied.
TlbEntry *entry = lookup(vaddr);
if (!entry) {
#if FULL_SYSTEM
2009-04-09 07:21:27 +02:00
Fault fault = walker->start(tc, translation, req, mode);
if (timing || fault != NoFault) {
// This gets ignored in atomic mode.
delayedResponse = true;
return fault;
}
entry = lookup(vaddr);
assert(entry);
#else
DPRINTF(TLB, "Handling a TLB miss for "
"address %#x at pc %#x.\n",
ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors. This change is a low level and pervasive reorganization of how PCs are managed in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about, the PC and the NPC, and the lsb of the PC signaled whether or not you were in PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next micropc, x86 and ARM introduced variable length instruction sets, and ARM started to keep track of mode bits in the PC. Each CPU model handled PCs in its own custom way that needed to be updated individually to handle the new dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack, the complexity could be hidden in the ISA at the ISA implementation's expense. Areas like the branch predictor hadn't been updated to handle branch delay slots or micropcs, and it turns out that had introduced a significant (10s of percent) performance bug in SPARC and to a lesser extend MIPS. Rather than perpetuate the problem by reworking O3 again to handle the PC features needed by x86, this change was introduced to rework PC handling in a more modular, transparent, and hopefully efficient way. PC type: Rather than having the superset of all possible elements of PC state declared in each of the CPU models, each ISA defines its own PCState type which has exactly the elements it needs. A cross product of canned PCState classes are defined in the new "generic" ISA directory for ISAs with/without delay slots and microcode. These are either typedef-ed or subclassed by each ISA. To read or write this structure through a *Context, you use the new pcState() accessor which reads or writes depending on whether it has an argument. If you just want the address of the current or next instruction or the current micro PC, you can get those through read-only accessors on either the PCState type or the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the move away from readPC. That name is ambiguous since it's not clear whether or not it should be the actual address to fetch from, or if it should have extra bits in it like the PAL mode bit. Each class is free to define its own functions to get at whatever values it needs however it needs to to be used in ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the PC and into a separate field like ARM. These types can be reset to a particular pc (where npc = pc + sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as appropriate), printed, serialized, and compared. There is a branching() function which encapsulates code in the CPU models that checked if an instruction branched or not. Exactly what that means in the context of branch delay slots which can skip an instruction when not taken is ambiguous, and ideally this function and its uses can be eliminated. PCStates also generally know how to advance themselves in various ways depending on if they point at an instruction, a microop, or the last microop of a macroop. More on that later. Ideally, accessing all the PCs at once when setting them will improve performance of M5 even though more data needs to be moved around. This is because often all the PCs need to be manipulated together, and by getting them all at once you avoid multiple function calls. Also, the PCs of a particular thread will have spatial locality in the cache. Previously they were grouped by element in arrays which spread out accesses. Advancing the PC: The PCs were previously managed entirely by the CPU which had to know about PC semantics, try to figure out which dimension to increment the PC in, what to set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction with the PC type itself. Because most of the information about how to increment the PC (mainly what type of instruction it refers to) is contained in the instruction object, a new advancePC virtual function was added to the StaticInst class. Subclasses provide an implementation that moves around the right element of the PC with a minimal amount of decision making. In ISAs like Alpha, the instructions always simply assign NPC to PC without having to worry about micropcs, nnpcs, etc. The added cost of a virtual function call should be outweighed by not having to figure out as much about what to do with the PCs and mucking around with the extra elements. One drawback of making the StaticInsts advance the PC is that you have to actually have one to advance the PC. This would, superficially, seem to require decoding an instruction before fetch could advance. This is, as far as I can tell, realistic. fetch would advance through memory addresses, not PCs, perhaps predicting new memory addresses using existing ones. More sophisticated decisions about control flow would be made later on, after the instruction was decoded, and handed back to fetch. If branching needs to happen, some amount of decoding needs to happen to see that it's a branch, what the target is, etc. This could get a little more complicated if that gets done by the predecoder, but I'm choosing to ignore that for now. Variable length instructions: To handle variable length instructions in x86 and ARM, the predecoder now takes in the current PC by reference to the getExtMachInst function. It can modify the PC however it needs to (by setting NPC to be the PC + instruction length, for instance). This could be improved since the CPU doesn't know if the PC was modified and always has to write it back. ISA parser: To support the new API, all PC related operand types were removed from the parser and replaced with a PCState type. There are two warts on this implementation. First, as with all the other operand types, the PCState still has to have a valid operand type even though it doesn't use it. Second, using syntax like PCS.npc(target) doesn't work for two reasons, this looks like the syntax for operand type overriding, and the parser can't figure out if you're reading or writing. Instructions that use the PCS operand (which I've consistently called it) need to first read it into a local variable, manipulate it, and then write it back out. Return address stack: The return address stack needed a little extra help because, in the presence of branch delay slots, it has to merge together elements of the return PC and the call PC. To handle that, a buildRetPC utility function was added. There are basically only two versions in all the ISAs, but it didn't seem short enough to put into the generic ISA directory. Also, the branch predictor code in O3 and InOrder were adjusted so that they always store the PC of the actual call instruction in the RAS, not the next PC. If the call instruction is a microop, the next PC refers to the next microop in the same macroop which is probably not desirable. The buildRetPC function advances the PC intelligently to the next macroop (in an ISA specific way) so that that case works. Change in stats: There were no change in stats except in MIPS and SPARC in the O3 model. MIPS runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could likely be improved further by setting call/return instruction flags and taking advantage of the RAS. TODO: Add != operators to the PCState classes, defined trivially to be !(a==b). Smooth out places where PCs are split apart, passed around, and put back together later. I think this might happen in SPARC's fault code. Add ISA specific constructors that allow setting PC elements without calling a bunch of accessors. Try to eliminate the need for the branching() function. Factor out Alpha's PAL mode pc bit into a separate flag field, and eliminate places where it's blindly masked out or tested in the PC.
2010-10-31 08:07:20 +01:00
vaddr, tc->instAddr());
Process *p = tc->getProcessPtr();
TlbEntry newEntry;
bool success = p->pTable->lookup(vaddr, newEntry);
if (!success && mode != Execute) {
// Check if we just need to grow the stack.
if (p->fixupStackFault(vaddr)) {
// If we did, lookup the entry for the new page.
success = p->pTable->lookup(vaddr, newEntry);
}
}
if (!success) {
return new PageFault(vaddr, true, mode, true, false);
} else {
Addr alignedVaddr = p->pTable->pageAlign(vaddr);
DPRINTF(TLB, "Mapping %#x to %#x\n", alignedVaddr,
newEntry.pageStart());
entry = insert(alignedVaddr, newEntry);
}
DPRINTF(TLB, "Miss was serviced.\n");
#endif
}
// Do paging protection checks.
bool inUser = (m5Reg.cpl == 3 &&
2009-02-25 19:18:58 +01:00
!(flags & (CPL0FlagBit << FlagShift)));
CR0 cr0 = tc->readMiscRegNoEffect(MISCREG_CR0);
bool badWrite = (!entry->writable && (inUser || cr0.wp));
if ((inUser && !entry->user) || (mode == Write && badWrite)) {
2009-02-25 19:18:58 +01:00
// The page must have been present to get into the TLB in
// the first place. We'll assume the reserved bits are
// fine even though we're not checking them.
2009-04-09 07:21:27 +02:00
return new PageFault(vaddr, true, mode, inUser, false);
2009-02-25 19:18:58 +01:00
}
if (storeCheck && badWrite) {
// This would fault if this were a write, so return a page
// fault that reflects that happening.
return new PageFault(vaddr, true, Write, inUser, false);
}
2009-02-25 19:18:58 +01:00
DPRINTF(TLB, "Entry found with paddr %#x, "
"doing protection checks.\n", entry->paddr);
Addr paddr = entry->paddr | (vaddr & (entry->size-1));
DPRINTF(TLB, "Translated %#x -> %#x.\n", vaddr, paddr);
req->setPaddr(paddr);
if (entry->uncacheable)
req->setFlags(Request::UNCACHEABLE);
} else {
//Use the address which already has segmentation applied.
DPRINTF(TLB, "Paging disabled.\n");
DPRINTF(TLB, "Translated %#x -> %#x.\n", vaddr, vaddr);
req->setPaddr(vaddr);
}
} else {
// Real mode
DPRINTF(TLB, "In real mode.\n");
DPRINTF(TLB, "Translated %#x -> %#x.\n", vaddr, vaddr);
req->setPaddr(vaddr);
}
// Check for an access to the local APIC
#if FULL_SYSTEM
LocalApicBase localApicBase = tc->readMiscRegNoEffect(MISCREG_APIC_BASE);
Addr baseAddr = localApicBase.base * PageBytes;
Addr paddr = req->getPaddr();
if (baseAddr <= paddr && baseAddr + PageBytes > paddr) {
// The Intel developer's manuals say the below restrictions apply,
// but the linux kernel, because of a compiler optimization, breaks
// them.
/*
// Check alignment
if (paddr & ((32/8) - 1))
return new GeneralProtection(0);
// Check access size
if (req->getSize() != (32/8))
return new GeneralProtection(0);
*/
// Force the access to be uncacheable.
req->setFlags(Request::UNCACHEABLE);
req->setPaddr(x86LocalAPICAddress(tc->contextId(), paddr - baseAddr));
}
#endif
return NoFault;
};
Fault
2009-04-09 07:21:27 +02:00
TLB::translateAtomic(RequestPtr req, ThreadContext *tc, Mode mode)
{
bool delayedResponse;
2009-04-09 07:21:27 +02:00
return TLB::translate(req, tc, NULL, mode, delayedResponse, false);
}
void
TLB::translateTiming(RequestPtr req, ThreadContext *tc,
2009-04-09 07:21:27 +02:00
Translation *translation, Mode mode)
{
bool delayedResponse;
assert(translation);
2009-04-09 07:21:27 +02:00
Fault fault =
TLB::translate(req, tc, translation, mode, delayedResponse, true);
if (!delayedResponse)
2009-04-09 07:21:27 +02:00
translation->finish(fault, req, tc, mode);
}
#if FULL_SYSTEM
Tick
TLB::doMmuRegRead(ThreadContext *tc, Packet *pkt)
{
return tc->getCpuPtr()->ticks(1);
}
Tick
TLB::doMmuRegWrite(ThreadContext *tc, Packet *pkt)
{
return tc->getCpuPtr()->ticks(1);
}
Walker *
TLB::getWalker()
{
return walker;
}
#endif
void
TLB::serialize(std::ostream &os)
{
}
void
TLB::unserialize(Checkpoint *cp, const std::string &section)
{
}
} // namespace X86ISA
X86ISA::TLB *
X86TLBParams::create()
{
return new X86ISA::TLB(this);
}