gem5/src/arch/arm/isa.cc

2009 lines
76 KiB
C++
Raw Normal View History

/*
* Copyright (c) 2010-2016 ARM Limited
* 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
* Ali Saidi
*/
#include "arch/arm/isa.hh"
#include "arch/arm/pmu.hh"
#include "arch/arm/system.hh"
#include "cpu/base.hh"
#include "cpu/checker/cpu.hh"
#include "debug/Arm.hh"
#include "debug/MiscRegs.hh"
#include "dev/arm/generic_timer.hh"
#include "params/ArmISA.hh"
#include "sim/faults.hh"
#include "sim/stat_control.hh"
#include "sim/system.hh"
namespace ArmISA
{
/**
* Some registers alias with others, and therefore need to be translated.
* For each entry:
* The first value is the misc register that is to be looked up
* the second value is the lower part of the translation
* the third the upper part
* aligned with reference documentation ARM DDI 0487A.i pp 1540-1543
*/
const struct ISA::MiscRegInitializerEntry
ISA::MiscRegSwitch[] = {
{MISCREG_ACTLR_EL1, {MISCREG_ACTLR_NS, 0}},
{MISCREG_AFSR0_EL1, {MISCREG_ADFSR_NS, 0}},
{MISCREG_AFSR1_EL1, {MISCREG_AIFSR_NS, 0}},
{MISCREG_AMAIR_EL1, {MISCREG_AMAIR0_NS, MISCREG_AMAIR1_NS}},
{MISCREG_CONTEXTIDR_EL1, {MISCREG_CONTEXTIDR_NS, 0}},
{MISCREG_CPACR_EL1, {MISCREG_CPACR, 0}},
{MISCREG_CSSELR_EL1, {MISCREG_CSSELR_NS, 0}},
{MISCREG_DACR32_EL2, {MISCREG_DACR_NS, 0}},
{MISCREG_FAR_EL1, {MISCREG_DFAR_NS, MISCREG_IFAR_NS}},
// ESR_EL1 -> DFSR
{MISCREG_HACR_EL2, {MISCREG_HACR, 0}},
{MISCREG_ACTLR_EL2, {MISCREG_HACTLR, 0}},
{MISCREG_AFSR0_EL2, {MISCREG_HADFSR, 0}},
{MISCREG_AFSR1_EL2, {MISCREG_HAIFSR, 0}},
{MISCREG_AMAIR_EL2, {MISCREG_HAMAIR0, MISCREG_HAMAIR1}},
{MISCREG_CPTR_EL2, {MISCREG_HCPTR, 0}},
{MISCREG_HCR_EL2, {MISCREG_HCR, 0 /*MISCREG_HCR2*/}},
{MISCREG_MDCR_EL2, {MISCREG_HDCR, 0}},
{MISCREG_FAR_EL2, {MISCREG_HDFAR, MISCREG_HIFAR}},
{MISCREG_MAIR_EL2, {MISCREG_HMAIR0, MISCREG_HMAIR1}},
{MISCREG_HPFAR_EL2, {MISCREG_HPFAR, 0}},
{MISCREG_SCTLR_EL2, {MISCREG_HSCTLR, 0}},
{MISCREG_ESR_EL2, {MISCREG_HSR, 0}},
{MISCREG_HSTR_EL2, {MISCREG_HSTR, 0}},
{MISCREG_TCR_EL2, {MISCREG_HTCR, 0}},
{MISCREG_TPIDR_EL2, {MISCREG_HTPIDR, 0}},
{MISCREG_TTBR0_EL2, {MISCREG_HTTBR, 0}},
{MISCREG_VBAR_EL2, {MISCREG_HVBAR, 0}},
{MISCREG_IFSR32_EL2, {MISCREG_IFSR_NS, 0}},
{MISCREG_MAIR_EL1, {MISCREG_PRRR_NS, MISCREG_NMRR_NS}},
{MISCREG_PAR_EL1, {MISCREG_PAR_NS, 0}},
// RMR_EL1 -> RMR
// RMR_EL2 -> HRMR
{MISCREG_SCTLR_EL1, {MISCREG_SCTLR_NS, 0}},
{MISCREG_SDER32_EL3, {MISCREG_SDER, 0}},
{MISCREG_TPIDR_EL1, {MISCREG_TPIDRPRW_NS, 0}},
{MISCREG_TPIDRRO_EL0, {MISCREG_TPIDRURO_NS, 0}},
{MISCREG_TPIDR_EL0, {MISCREG_TPIDRURW_NS, 0}},
{MISCREG_TCR_EL1, {MISCREG_TTBCR_NS, 0}},
{MISCREG_TTBR0_EL1, {MISCREG_TTBR0_NS, 0}},
{MISCREG_TTBR1_EL1, {MISCREG_TTBR1_NS, 0}},
{MISCREG_VBAR_EL1, {MISCREG_VBAR_NS, 0}},
{MISCREG_VMPIDR_EL2, {MISCREG_VMPIDR, 0}},
{MISCREG_VPIDR_EL2, {MISCREG_VPIDR, 0}},
{MISCREG_VTCR_EL2, {MISCREG_VTCR, 0}},
{MISCREG_VTTBR_EL2, {MISCREG_VTTBR, 0}},
{MISCREG_CNTFRQ_EL0, {MISCREG_CNTFRQ, 0}},
{MISCREG_CNTHCTL_EL2, {MISCREG_CNTHCTL, 0}},
{MISCREG_CNTHP_CTL_EL2, {MISCREG_CNTHP_CTL, 0}},
{MISCREG_CNTHP_CVAL_EL2, {MISCREG_CNTHP_CVAL, 0}}, /* 64b */
{MISCREG_CNTHP_TVAL_EL2, {MISCREG_CNTHP_TVAL, 0}},
{MISCREG_CNTKCTL_EL1, {MISCREG_CNTKCTL, 0}},
{MISCREG_CNTP_CTL_EL0, {MISCREG_CNTP_CTL_NS, 0}},
{MISCREG_CNTP_CVAL_EL0, {MISCREG_CNTP_CVAL_NS, 0}}, /* 64b */
{MISCREG_CNTP_TVAL_EL0, {MISCREG_CNTP_TVAL_NS, 0}},
{MISCREG_CNTPCT_EL0, {MISCREG_CNTPCT, 0}}, /* 64b */
{MISCREG_CNTV_CTL_EL0, {MISCREG_CNTV_CTL, 0}},
{MISCREG_CNTV_CVAL_EL0, {MISCREG_CNTV_CVAL, 0}}, /* 64b */
{MISCREG_CNTV_TVAL_EL0, {MISCREG_CNTV_TVAL, 0}},
{MISCREG_CNTVCT_EL0, {MISCREG_CNTVCT, 0}}, /* 64b */
{MISCREG_CNTVOFF_EL2, {MISCREG_CNTVOFF, 0}}, /* 64b */
{MISCREG_DBGAUTHSTATUS_EL1, {MISCREG_DBGAUTHSTATUS, 0}},
{MISCREG_DBGBCR0_EL1, {MISCREG_DBGBCR0, 0}},
{MISCREG_DBGBCR1_EL1, {MISCREG_DBGBCR1, 0}},
{MISCREG_DBGBCR2_EL1, {MISCREG_DBGBCR2, 0}},
{MISCREG_DBGBCR3_EL1, {MISCREG_DBGBCR3, 0}},
{MISCREG_DBGBCR4_EL1, {MISCREG_DBGBCR4, 0}},
{MISCREG_DBGBCR5_EL1, {MISCREG_DBGBCR5, 0}},
{MISCREG_DBGBVR0_EL1, {MISCREG_DBGBVR0, 0 /* MISCREG_DBGBXVR0 */}},
{MISCREG_DBGBVR1_EL1, {MISCREG_DBGBVR1, 0 /* MISCREG_DBGBXVR1 */}},
{MISCREG_DBGBVR2_EL1, {MISCREG_DBGBVR2, 0 /* MISCREG_DBGBXVR2 */}},
{MISCREG_DBGBVR3_EL1, {MISCREG_DBGBVR3, 0 /* MISCREG_DBGBXVR3 */}},
{MISCREG_DBGBVR4_EL1, {MISCREG_DBGBVR4, MISCREG_DBGBXVR4}},
{MISCREG_DBGBVR5_EL1, {MISCREG_DBGBVR5, MISCREG_DBGBXVR5}},
{MISCREG_DBGCLAIMSET_EL1, {MISCREG_DBGCLAIMSET, 0}},
{MISCREG_DBGCLAIMCLR_EL1, {MISCREG_DBGCLAIMCLR, 0}},
// DBGDTR_EL0 -> DBGDTR{R or T}Xint
// DBGDTRRX_EL0 -> DBGDTRRXint
// DBGDTRTX_EL0 -> DBGDTRRXint
{MISCREG_DBGPRCR_EL1, {MISCREG_DBGPRCR, 0}},
{MISCREG_DBGVCR32_EL2, {MISCREG_DBGVCR, 0}},
{MISCREG_DBGWCR0_EL1, {MISCREG_DBGWCR0, 0}},
{MISCREG_DBGWCR1_EL1, {MISCREG_DBGWCR1, 0}},
{MISCREG_DBGWCR2_EL1, {MISCREG_DBGWCR2, 0}},
{MISCREG_DBGWCR3_EL1, {MISCREG_DBGWCR3, 0}},
{MISCREG_DBGWVR0_EL1, {MISCREG_DBGWVR0, 0}},
{MISCREG_DBGWVR1_EL1, {MISCREG_DBGWVR1, 0}},
{MISCREG_DBGWVR2_EL1, {MISCREG_DBGWVR2, 0}},
{MISCREG_DBGWVR3_EL1, {MISCREG_DBGWVR3, 0}},
{MISCREG_ID_DFR0_EL1, {MISCREG_ID_DFR0, 0}},
{MISCREG_MDCCSR_EL0, {MISCREG_DBGDSCRint, 0}},
{MISCREG_MDRAR_EL1, {MISCREG_DBGDRAR, 0}},
{MISCREG_MDSCR_EL1, {MISCREG_DBGDSCRext, 0}},
{MISCREG_OSDLR_EL1, {MISCREG_DBGOSDLR, 0}},
{MISCREG_OSDTRRX_EL1, {MISCREG_DBGDTRRXext, 0}},
{MISCREG_OSDTRTX_EL1, {MISCREG_DBGDTRTXext, 0}},
{MISCREG_OSECCR_EL1, {MISCREG_DBGOSECCR, 0}},
{MISCREG_OSLAR_EL1, {MISCREG_DBGOSLAR, 0}},
{MISCREG_OSLSR_EL1, {MISCREG_DBGOSLSR, 0}},
{MISCREG_PMCCNTR_EL0, {MISCREG_PMCCNTR, 0}},
{MISCREG_PMCEID0_EL0, {MISCREG_PMCEID0, 0}},
{MISCREG_PMCEID1_EL0, {MISCREG_PMCEID1, 0}},
{MISCREG_PMCNTENSET_EL0, {MISCREG_PMCNTENSET, 0}},
{MISCREG_PMCNTENCLR_EL0, {MISCREG_PMCNTENCLR, 0}},
{MISCREG_PMCR_EL0, {MISCREG_PMCR, 0}},
/* {MISCREG_PMEVCNTR0_EL0, {MISCREG_PMEVCNTR0, 0}},
{MISCREG_PMEVCNTR1_EL0, {MISCREG_PMEVCNTR1, 0}},
{MISCREG_PMEVCNTR2_EL0, {MISCREG_PMEVCNTR2, 0}},
{MISCREG_PMEVCNTR3_EL0, {MISCREG_PMEVCNTR3, 0}},
{MISCREG_PMEVCNTR4_EL0, {MISCREG_PMEVCNTR4, 0}},
{MISCREG_PMEVCNTR5_EL0, {MISCREG_PMEVCNTR5, 0}},
{MISCREG_PMEVTYPER0_EL0, {MISCREG_PMEVTYPER0, 0}},
{MISCREG_PMEVTYPER1_EL0, {MISCREG_PMEVTYPER1, 0}},
{MISCREG_PMEVTYPER2_EL0, {MISCREG_PMEVTYPER2, 0}},
{MISCREG_PMEVTYPER3_EL0, {MISCREG_PMEVTYPER3, 0}},
{MISCREG_PMEVTYPER4_EL0, {MISCREG_PMEVTYPER4, 0}},
{MISCREG_PMEVTYPER5_EL0, {MISCREG_PMEVTYPER5, 0}}, */
{MISCREG_PMINTENCLR_EL1, {MISCREG_PMINTENCLR, 0}},
{MISCREG_PMINTENSET_EL1, {MISCREG_PMINTENSET, 0}},
// {MISCREG_PMOVSCLR_EL0, {MISCREG_PMOVSCLR, 0}},
{MISCREG_PMOVSSET_EL0, {MISCREG_PMOVSSET, 0}},
{MISCREG_PMSELR_EL0, {MISCREG_PMSELR, 0}},
{MISCREG_PMSWINC_EL0, {MISCREG_PMSWINC, 0}},
{MISCREG_PMUSERENR_EL0, {MISCREG_PMUSERENR, 0}},
{MISCREG_PMXEVCNTR_EL0, {MISCREG_PMXEVCNTR, 0}},
{MISCREG_PMXEVTYPER_EL0, {MISCREG_PMXEVTYPER, 0}},
// from ARM DDI 0487A.i, template text
// "AArch64 System register ___ can be mapped to
// AArch32 System register ___, but this is not
// architecturally mandated."
{MISCREG_SCR_EL3, {MISCREG_SCR, 0}}, // D7-2005
// MDCR_EL3 -> SDCR, D7-2108 (the latter is unimpl. in gem5)
{MISCREG_SPSR_EL1, {MISCREG_SPSR_SVC, 0}}, // C5.2.17 SPSR_EL1
{MISCREG_SPSR_EL2, {MISCREG_SPSR_HYP, 0}}, // C5.2.18 SPSR_EL2
{MISCREG_SPSR_EL3, {MISCREG_SPSR_MON, 0}}, // C5.2.19 SPSR_EL3
};
ISA::ISA(Params *p)
: SimObject(p),
system(NULL),
_decoderFlavour(p->decoderFlavour),
pmu(p->pmu),
lookUpMiscReg(NUM_MISCREGS, {0,0})
{
miscRegs[MISCREG_SCTLR_RST] = 0;
// Hook up a dummy device if we haven't been configured with a
// real PMU. By using a dummy device, we don't need to check that
// the PMU exist every time we try to access a PMU register.
if (!pmu)
pmu = &dummyDevice;
// Give all ISA devices a pointer to this ISA
pmu->setISA(this);
system = dynamic_cast<ArmSystem *>(p->system);
// Cache system-level properties
if (FullSystem && system) {
highestELIs64 = system->highestELIs64();
haveSecurity = system->haveSecurity();
haveLPAE = system->haveLPAE();
haveVirtualization = system->haveVirtualization();
haveLargeAsid64 = system->haveLargeAsid64();
physAddrRange64 = system->physAddrRange64();
} else {
highestELIs64 = true; // ArmSystem::highestELIs64 does the same
haveSecurity = haveLPAE = haveVirtualization = false;
haveLargeAsid64 = false;
physAddrRange64 = 32; // dummy value
}
/** Fill in the miscReg translation table */
for (auto sw : MiscRegSwitch) {
lookUpMiscReg[sw.index] = sw.entry;
}
preUnflattenMiscReg();
clear();
}
const ArmISAParams *
ISA::params() const
{
return dynamic_cast<const Params *>(_params);
}
void
ISA::clear()
{
const Params *p(params());
SCTLR sctlr_rst = miscRegs[MISCREG_SCTLR_RST];
memset(miscRegs, 0, sizeof(miscRegs));
// Initialize configurable default values
miscRegs[MISCREG_MIDR] = p->midr;
miscRegs[MISCREG_MIDR_EL1] = p->midr;
miscRegs[MISCREG_VPIDR] = p->midr;
if (FullSystem && system->highestELIs64()) {
// Initialize AArch64 state
clear64(p);
return;
}
// Initialize AArch32 state...
CPSR cpsr = 0;
cpsr.mode = MODE_USER;
miscRegs[MISCREG_CPSR] = cpsr;
updateRegMap(cpsr);
SCTLR sctlr = 0;
sctlr.te = (bool) sctlr_rst.te;
sctlr.nmfi = (bool) sctlr_rst.nmfi;
sctlr.v = (bool) sctlr_rst.v;
sctlr.u = 1;
sctlr.xp = 1;
sctlr.rao2 = 1;
sctlr.rao3 = 1;
sctlr.rao4 = 0xf; // SCTLR[6:3]
sctlr.uci = 1;
sctlr.dze = 1;
miscRegs[MISCREG_SCTLR_NS] = sctlr;
miscRegs[MISCREG_SCTLR_RST] = sctlr_rst;
miscRegs[MISCREG_HCPTR] = 0;
// Start with an event in the mailbox
miscRegs[MISCREG_SEV_MAILBOX] = 1;
// Separate Instruction and Data TLBs
miscRegs[MISCREG_TLBTR] = 1;
MVFR0 mvfr0 = 0;
mvfr0.advSimdRegisters = 2;
mvfr0.singlePrecision = 2;
mvfr0.doublePrecision = 2;
mvfr0.vfpExceptionTrapping = 0;
mvfr0.divide = 1;
mvfr0.squareRoot = 1;
mvfr0.shortVectors = 1;
mvfr0.roundingModes = 1;
miscRegs[MISCREG_MVFR0] = mvfr0;
MVFR1 mvfr1 = 0;
mvfr1.flushToZero = 1;
mvfr1.defaultNaN = 1;
mvfr1.advSimdLoadStore = 1;
mvfr1.advSimdInteger = 1;
mvfr1.advSimdSinglePrecision = 1;
mvfr1.advSimdHalfPrecision = 1;
mvfr1.vfpHalfPrecision = 1;
miscRegs[MISCREG_MVFR1] = mvfr1;
// Reset values of PRRR and NMRR are implementation dependent
// @todo: PRRR and NMRR in secure state?
miscRegs[MISCREG_PRRR_NS] =
(1 << 19) | // 19
(0 << 18) | // 18
(0 << 17) | // 17
(1 << 16) | // 16
(2 << 14) | // 15:14
(0 << 12) | // 13:12
(2 << 10) | // 11:10
(2 << 8) | // 9:8
(2 << 6) | // 7:6
(2 << 4) | // 5:4
(1 << 2) | // 3:2
0; // 1:0
miscRegs[MISCREG_NMRR_NS] =
(1 << 30) | // 31:30
(0 << 26) | // 27:26
(0 << 24) | // 25:24
(3 << 22) | // 23:22
(2 << 20) | // 21:20
(0 << 18) | // 19:18
(0 << 16) | // 17:16
(1 << 14) | // 15:14
(0 << 12) | // 13:12
(2 << 10) | // 11:10
(0 << 8) | // 9:8
(3 << 6) | // 7:6
(2 << 4) | // 5:4
(0 << 2) | // 3:2
0; // 1:0
miscRegs[MISCREG_CPACR] = 0;
miscRegs[MISCREG_ID_MMFR0] = p->id_mmfr0;
miscRegs[MISCREG_ID_MMFR1] = p->id_mmfr1;
miscRegs[MISCREG_ID_MMFR2] = p->id_mmfr2;
miscRegs[MISCREG_ID_MMFR3] = p->id_mmfr3;
miscRegs[MISCREG_ID_ISAR0] = p->id_isar0;
miscRegs[MISCREG_ID_ISAR1] = p->id_isar1;
miscRegs[MISCREG_ID_ISAR2] = p->id_isar2;
miscRegs[MISCREG_ID_ISAR3] = p->id_isar3;
miscRegs[MISCREG_ID_ISAR4] = p->id_isar4;
miscRegs[MISCREG_ID_ISAR5] = p->id_isar5;
miscRegs[MISCREG_FPSID] = p->fpsid;
if (haveLPAE) {
TTBCR ttbcr = miscRegs[MISCREG_TTBCR_NS];
ttbcr.eae = 0;
miscRegs[MISCREG_TTBCR_NS] = ttbcr;
// Enforce consistency with system-level settings
miscRegs[MISCREG_ID_MMFR0] = (miscRegs[MISCREG_ID_MMFR0] & ~0xf) | 0x5;
}
if (haveSecurity) {
miscRegs[MISCREG_SCTLR_S] = sctlr;
miscRegs[MISCREG_SCR] = 0;
miscRegs[MISCREG_VBAR_S] = 0;
} else {
// we're always non-secure
miscRegs[MISCREG_SCR] = 1;
}
//XXX We need to initialize the rest of the state.
}
void
ISA::clear64(const ArmISAParams *p)
{
CPSR cpsr = 0;
Addr rvbar = system->resetAddr64();
switch (system->highestEL()) {
// Set initial EL to highest implemented EL using associated stack
// pointer (SP_ELx); set RVBAR_ELx to implementation defined reset
// value
case EL3:
cpsr.mode = MODE_EL3H;
miscRegs[MISCREG_RVBAR_EL3] = rvbar;
break;
case EL2:
cpsr.mode = MODE_EL2H;
miscRegs[MISCREG_RVBAR_EL2] = rvbar;
break;
case EL1:
cpsr.mode = MODE_EL1H;
miscRegs[MISCREG_RVBAR_EL1] = rvbar;
break;
default:
panic("Invalid highest implemented exception level");
break;
}
// Initialize rest of CPSR
cpsr.daif = 0xf; // Mask all interrupts
cpsr.ss = 0;
cpsr.il = 0;
miscRegs[MISCREG_CPSR] = cpsr;
updateRegMap(cpsr);
// Initialize other control registers
miscRegs[MISCREG_MPIDR_EL1] = 0x80000000;
if (haveSecurity) {
miscRegs[MISCREG_SCTLR_EL3] = 0x30c50830;
miscRegs[MISCREG_SCR_EL3] = 0x00000030; // RES1 fields
} else if (haveVirtualization) {
// also MISCREG_SCTLR_EL2 (by mapping)
miscRegs[MISCREG_HSCTLR] = 0x30c50830;
} else {
// also MISCREG_SCTLR_EL1 (by mapping)
miscRegs[MISCREG_SCTLR_NS] = 0x30d00800 | 0x00050030; // RES1 | init
// Always non-secure
miscRegs[MISCREG_SCR_EL3] = 1;
}
// Initialize configurable id registers
miscRegs[MISCREG_ID_AA64AFR0_EL1] = p->id_aa64afr0_el1;
miscRegs[MISCREG_ID_AA64AFR1_EL1] = p->id_aa64afr1_el1;
miscRegs[MISCREG_ID_AA64DFR0_EL1] =
(p->id_aa64dfr0_el1 & 0xfffffffffffff0ffULL) |
(p->pmu ? 0x0000000000000100ULL : 0); // Enable PMUv3
miscRegs[MISCREG_ID_AA64DFR1_EL1] = p->id_aa64dfr1_el1;
miscRegs[MISCREG_ID_AA64ISAR0_EL1] = p->id_aa64isar0_el1;
miscRegs[MISCREG_ID_AA64ISAR1_EL1] = p->id_aa64isar1_el1;
miscRegs[MISCREG_ID_AA64MMFR0_EL1] = p->id_aa64mmfr0_el1;
miscRegs[MISCREG_ID_AA64MMFR1_EL1] = p->id_aa64mmfr1_el1;
miscRegs[MISCREG_ID_DFR0_EL1] =
(p->pmu ? 0x03000000ULL : 0); // Enable PMUv3
miscRegs[MISCREG_ID_DFR0] = miscRegs[MISCREG_ID_DFR0_EL1];
// Enforce consistency with system-level settings...
// EL3
miscRegs[MISCREG_ID_AA64PFR0_EL1] = insertBits(
miscRegs[MISCREG_ID_AA64PFR0_EL1], 15, 12,
haveSecurity ? 0x2 : 0x0);
// EL2
miscRegs[MISCREG_ID_AA64PFR0_EL1] = insertBits(
miscRegs[MISCREG_ID_AA64PFR0_EL1], 11, 8,
haveVirtualization ? 0x2 : 0x0);
// Large ASID support
miscRegs[MISCREG_ID_AA64MMFR0_EL1] = insertBits(
miscRegs[MISCREG_ID_AA64MMFR0_EL1], 7, 4,
haveLargeAsid64 ? 0x2 : 0x0);
// Physical address size
miscRegs[MISCREG_ID_AA64MMFR0_EL1] = insertBits(
miscRegs[MISCREG_ID_AA64MMFR0_EL1], 3, 0,
encodePhysAddrRange64(physAddrRange64));
}
MiscReg
ISA::readMiscRegNoEffect(int misc_reg) const
{
assert(misc_reg < NumMiscRegs);
auto regs = getMiscIndices(misc_reg);
int lower = regs.first, upper = regs.second;
return !upper ? miscRegs[lower] : ((miscRegs[lower] & mask(32))
|(miscRegs[upper] << 32));
}
MiscReg
ISA::readMiscReg(int misc_reg, ThreadContext *tc)
{
CPSR cpsr = 0;
PCState pc = 0;
SCR scr = 0;
if (misc_reg == MISCREG_CPSR) {
cpsr = miscRegs[misc_reg];
pc = tc->pcState();
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
cpsr.j = pc.jazelle() ? 1 : 0;
cpsr.t = pc.thumb() ? 1 : 0;
return cpsr;
}
#ifndef NDEBUG
if (!miscRegInfo[misc_reg][MISCREG_IMPLEMENTED]) {
if (miscRegInfo[misc_reg][MISCREG_WARN_NOT_FAIL])
warn("Unimplemented system register %s read.\n",
miscRegName[misc_reg]);
else
panic("Unimplemented system register %s read.\n",
miscRegName[misc_reg]);
}
#endif
switch (unflattenMiscReg(misc_reg)) {
case MISCREG_HCR:
{
if (!haveVirtualization)
return 0;
else
return readMiscRegNoEffect(MISCREG_HCR);
}
case MISCREG_CPACR:
{
const uint32_t ones = (uint32_t)(-1);
CPACR cpacrMask = 0;
// Only cp10, cp11, and ase are implemented, nothing else should
// be readable? (straight copy from the write code)
cpacrMask.cp10 = ones;
cpacrMask.cp11 = ones;
cpacrMask.asedis = ones;
// Security Extensions may limit the readability of CPACR
if (haveSecurity) {
scr = readMiscRegNoEffect(MISCREG_SCR);
cpsr = readMiscRegNoEffect(MISCREG_CPSR);
if (scr.ns && (cpsr.mode != MODE_MON)) {
NSACR nsacr = readMiscRegNoEffect(MISCREG_NSACR);
// NB: Skipping the full loop, here
if (!nsacr.cp10) cpacrMask.cp10 = 0;
if (!nsacr.cp11) cpacrMask.cp11 = 0;
}
}
MiscReg val = readMiscRegNoEffect(MISCREG_CPACR);
val &= cpacrMask;
DPRINTF(MiscRegs, "Reading misc reg %s: %#x\n",
miscRegName[misc_reg], val);
return val;
}
case MISCREG_MPIDR:
cpsr = readMiscRegNoEffect(MISCREG_CPSR);
scr = readMiscRegNoEffect(MISCREG_SCR);
if ((cpsr.mode == MODE_HYP) || inSecureState(scr, cpsr)) {
return getMPIDR(system, tc);
} else {
return readMiscReg(MISCREG_VMPIDR, tc);
}
break;
case MISCREG_MPIDR_EL1:
// @todo in the absence of v8 virtualization support just return MPIDR_EL1
return getMPIDR(system, tc) & 0xffffffff;
case MISCREG_VMPIDR:
// top bit defined as RES1
return readMiscRegNoEffect(misc_reg) | 0x80000000;
case MISCREG_ID_AFR0: // not implemented, so alias MIDR
case MISCREG_REVIDR: // not implemented, so alias MIDR
case MISCREG_MIDR:
cpsr = readMiscRegNoEffect(MISCREG_CPSR);
scr = readMiscRegNoEffect(MISCREG_SCR);
if ((cpsr.mode == MODE_HYP) || inSecureState(scr, cpsr)) {
return readMiscRegNoEffect(misc_reg);
} else {
return readMiscRegNoEffect(MISCREG_VPIDR);
}
break;
case MISCREG_JOSCR: // Jazelle trivial implementation, RAZ/WI
case MISCREG_JMCR: // Jazelle trivial implementation, RAZ/WI
case MISCREG_JIDR: // Jazelle trivial implementation, RAZ/WI
case MISCREG_AIDR: // AUX ID set to 0
case MISCREG_TCMTR: // No TCM's
return 0;
case MISCREG_CLIDR:
warn_once("The clidr register always reports 0 caches.\n");
warn_once("clidr LoUIS field of 0b001 to match current "
"ARM implementations.\n");
return 0x00200000;
case MISCREG_CCSIDR:
warn_once("The ccsidr register isn't implemented and "
"always reads as 0.\n");
break;
case MISCREG_CTR:
{
//all caches have the same line size in gem5
//4 byte words in ARM
unsigned lineSizeWords =
tc->getSystemPtr()->cacheLineSize() / 4;
unsigned log2LineSizeWords = 0;
while (lineSizeWords >>= 1) {
++log2LineSizeWords;
}
CTR ctr = 0;
//log2 of minimun i-cache line size (words)
ctr.iCacheLineSize = log2LineSizeWords;
//b11 - gem5 uses pipt
ctr.l1IndexPolicy = 0x3;
//log2 of minimum d-cache line size (words)
ctr.dCacheLineSize = log2LineSizeWords;
//log2 of max reservation size (words)
ctr.erg = log2LineSizeWords;
//log2 of max writeback size (words)
ctr.cwg = log2LineSizeWords;
//b100 - gem5 format is ARMv7
ctr.format = 0x4;
return ctr;
}
case MISCREG_ACTLR:
warn("Not doing anything for miscreg ACTLR\n");
break;
case MISCREG_PMXEVTYPER_PMCCFILTR:
case MISCREG_PMINTENSET_EL1 ... MISCREG_PMOVSSET_EL0:
case MISCREG_PMEVCNTR0_EL0 ... MISCREG_PMEVTYPER5_EL0:
case MISCREG_PMCR ... MISCREG_PMOVSSET:
return pmu->readMiscReg(misc_reg);
case MISCREG_CPSR_Q:
panic("shouldn't be reading this register seperately\n");
case MISCREG_FPSCR_QC:
return readMiscRegNoEffect(MISCREG_FPSCR) & ~FpscrQcMask;
case MISCREG_FPSCR_EXC:
return readMiscRegNoEffect(MISCREG_FPSCR) & ~FpscrExcMask;
case MISCREG_FPSR:
{
const uint32_t ones = (uint32_t)(-1);
FPSCR fpscrMask = 0;
fpscrMask.ioc = ones;
fpscrMask.dzc = ones;
fpscrMask.ofc = ones;
fpscrMask.ufc = ones;
fpscrMask.ixc = ones;
fpscrMask.idc = ones;
fpscrMask.qc = ones;
fpscrMask.v = ones;
fpscrMask.c = ones;
fpscrMask.z = ones;
fpscrMask.n = ones;
return readMiscRegNoEffect(MISCREG_FPSCR) & (uint32_t)fpscrMask;
}
case MISCREG_FPCR:
{
const uint32_t ones = (uint32_t)(-1);
FPSCR fpscrMask = 0;
fpscrMask.ioe = ones;
fpscrMask.dze = ones;
fpscrMask.ofe = ones;
fpscrMask.ufe = ones;
fpscrMask.ixe = ones;
fpscrMask.ide = ones;
fpscrMask.len = ones;
fpscrMask.stride = ones;
fpscrMask.rMode = ones;
fpscrMask.fz = ones;
fpscrMask.dn = ones;
fpscrMask.ahp = ones;
return readMiscRegNoEffect(MISCREG_FPSCR) & (uint32_t)fpscrMask;
}
case MISCREG_NZCV:
{
CPSR cpsr = 0;
cpsr.nz = tc->readCCReg(CCREG_NZ);
cpsr.c = tc->readCCReg(CCREG_C);
cpsr.v = tc->readCCReg(CCREG_V);
return cpsr;
}
case MISCREG_DAIF:
{
CPSR cpsr = 0;
cpsr.daif = (uint8_t) ((CPSR) miscRegs[MISCREG_CPSR]).daif;
return cpsr;
}
case MISCREG_SP_EL0:
{
return tc->readIntReg(INTREG_SP0);
}
case MISCREG_SP_EL1:
{
return tc->readIntReg(INTREG_SP1);
}
case MISCREG_SP_EL2:
{
return tc->readIntReg(INTREG_SP2);
}
case MISCREG_SPSEL:
{
return miscRegs[MISCREG_CPSR] & 0x1;
}
case MISCREG_CURRENTEL:
{
return miscRegs[MISCREG_CPSR] & 0xc;
}
case MISCREG_L2CTLR:
{
// mostly unimplemented, just set NumCPUs field from sim and return
L2CTLR l2ctlr = 0;
// b00:1CPU to b11:4CPUs
l2ctlr.numCPUs = tc->getSystemPtr()->numContexts() - 1;
return l2ctlr;
}
case MISCREG_DBGDIDR:
/* For now just implement the version number.
* ARMv7, v7.1 Debug architecture (0b0101 --> 0x5)
*/
return 0x5 << 16;
case MISCREG_DBGDSCRint:
return 0;
case MISCREG_ISR:
return tc->getCpuPtr()->getInterruptController(tc->threadId())->getISR(
readMiscRegNoEffect(MISCREG_HCR),
readMiscRegNoEffect(MISCREG_CPSR),
readMiscRegNoEffect(MISCREG_SCR));
case MISCREG_ISR_EL1:
return tc->getCpuPtr()->getInterruptController(tc->threadId())->getISR(
readMiscRegNoEffect(MISCREG_HCR_EL2),
readMiscRegNoEffect(MISCREG_CPSR),
readMiscRegNoEffect(MISCREG_SCR_EL3));
case MISCREG_DCZID_EL0:
return 0x04; // DC ZVA clear 64-byte chunks
case MISCREG_HCPTR:
{
MiscReg val = readMiscRegNoEffect(misc_reg);
// The trap bit associated with CP14 is defined as RAZ
val &= ~(1 << 14);
// If a CP bit in NSACR is 0 then the corresponding bit in
// HCPTR is RAO/WI
bool secure_lookup = haveSecurity &&
inSecureState(readMiscRegNoEffect(MISCREG_SCR),
readMiscRegNoEffect(MISCREG_CPSR));
if (!secure_lookup) {
MiscReg mask = readMiscRegNoEffect(MISCREG_NSACR);
val |= (mask ^ 0x7FFF) & 0xBFFF;
}
// Set the bits for unimplemented coprocessors to RAO/WI
val |= 0x33FF;
return (val);
}
case MISCREG_HDFAR: // alias for secure DFAR
return readMiscRegNoEffect(MISCREG_DFAR_S);
case MISCREG_HIFAR: // alias for secure IFAR
return readMiscRegNoEffect(MISCREG_IFAR_S);
case MISCREG_HVBAR: // bottom bits reserved
return readMiscRegNoEffect(MISCREG_HVBAR) & 0xFFFFFFE0;
case MISCREG_SCTLR:
return (readMiscRegNoEffect(misc_reg) & 0x72DD39FF) | 0x00C00818;
case MISCREG_SCTLR_EL1:
return (readMiscRegNoEffect(misc_reg) & 0x37DDDBBF) | 0x30D00800;
case MISCREG_SCTLR_EL2:
case MISCREG_SCTLR_EL3:
case MISCREG_HSCTLR:
return (readMiscRegNoEffect(misc_reg) & 0x32CD183F) | 0x30C50830;
case MISCREG_ID_PFR0:
// !ThumbEE | !Jazelle | Thumb | ARM
return 0x00000031;
case MISCREG_ID_PFR1:
{ // Timer | Virti | !M Profile | TrustZone | ARMv4
bool haveTimer = (system->getGenericTimer() != NULL);
return 0x00000001
| (haveSecurity ? 0x00000010 : 0x0)
| (haveVirtualization ? 0x00001000 : 0x0)
| (haveTimer ? 0x00010000 : 0x0);
}
case MISCREG_ID_AA64PFR0_EL1:
return 0x0000000000000002 // AArch{64,32} supported at EL0
| 0x0000000000000020 // EL1
| (haveVirtualization ? 0x0000000000000200 : 0) // EL2
| (haveSecurity ? 0x0000000000002000 : 0); // EL3
case MISCREG_ID_AA64PFR1_EL1:
return 0; // bits [63:0] RES0 (reserved for future use)
// Generic Timer registers
case MISCREG_CNTFRQ ... MISCREG_CNTHP_CTL:
case MISCREG_CNTPCT ... MISCREG_CNTHP_CVAL:
case MISCREG_CNTKCTL_EL1 ... MISCREG_CNTV_CVAL_EL0:
case MISCREG_CNTVOFF_EL2 ... MISCREG_CNTPS_CVAL_EL1:
return getGenericTimer(tc).readMiscReg(misc_reg);
default:
break;
}
return readMiscRegNoEffect(misc_reg);
}
void
ISA::setMiscRegNoEffect(int misc_reg, const MiscReg &val)
{
assert(misc_reg < NumMiscRegs);
auto regs = getMiscIndices(misc_reg);
int lower = regs.first, upper = regs.second;
if (upper > 0) {
miscRegs[lower] = bits(val, 31, 0);
miscRegs[upper] = bits(val, 63, 32);
DPRINTF(MiscRegs, "Writing to misc reg %d (%d:%d) : %#x\n",
misc_reg, lower, upper, val);
} else {
miscRegs[lower] = val;
DPRINTF(MiscRegs, "Writing to misc reg %d (%d) : %#x\n",
misc_reg, lower, val);
}
}
void
ISA::setMiscReg(int misc_reg, const MiscReg &val, ThreadContext *tc)
{
MiscReg newVal = val;
int x;
bool secure_lookup;
bool hyp;
System *sys;
ThreadContext *oc;
uint8_t target_el;
uint16_t asid;
SCR scr;
if (misc_reg == MISCREG_CPSR) {
updateRegMap(val);
CPSR old_cpsr = miscRegs[MISCREG_CPSR];
int old_mode = old_cpsr.mode;
CPSR cpsr = val;
if (old_mode != cpsr.mode) {
tc->getITBPtr()->invalidateMiscReg();
tc->getDTBPtr()->invalidateMiscReg();
}
DPRINTF(Arm, "Updating CPSR from %#x to %#x f:%d i:%d a:%d mode:%#x\n",
miscRegs[misc_reg], cpsr, cpsr.f, cpsr.i, cpsr.a, cpsr.mode);
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
PCState pc = tc->pcState();
pc.nextThumb(cpsr.t);
pc.nextJazelle(cpsr.j);
// Follow slightly different semantics if a CheckerCPU object
// is connected
CheckerCPU *checker = tc->getCheckerCpuPtr();
if (checker) {
tc->pcStateNoRecord(pc);
} else {
tc->pcState(pc);
}
} else {
#ifndef NDEBUG
if (!miscRegInfo[misc_reg][MISCREG_IMPLEMENTED]) {
if (miscRegInfo[misc_reg][MISCREG_WARN_NOT_FAIL])
warn("Unimplemented system register %s write with %#x.\n",
miscRegName[misc_reg], val);
else
panic("Unimplemented system register %s write with %#x.\n",
miscRegName[misc_reg], val);
}
#endif
switch (unflattenMiscReg(misc_reg)) {
case MISCREG_CPACR:
{
const uint32_t ones = (uint32_t)(-1);
CPACR cpacrMask = 0;
// Only cp10, cp11, and ase are implemented, nothing else should
// be writable
cpacrMask.cp10 = ones;
cpacrMask.cp11 = ones;
cpacrMask.asedis = ones;
// Security Extensions may limit the writability of CPACR
if (haveSecurity) {
scr = readMiscRegNoEffect(MISCREG_SCR);
CPSR cpsr = readMiscRegNoEffect(MISCREG_CPSR);
if (scr.ns && (cpsr.mode != MODE_MON)) {
NSACR nsacr = readMiscRegNoEffect(MISCREG_NSACR);
// NB: Skipping the full loop, here
if (!nsacr.cp10) cpacrMask.cp10 = 0;
if (!nsacr.cp11) cpacrMask.cp11 = 0;
}
}
MiscReg old_val = readMiscRegNoEffect(MISCREG_CPACR);
newVal &= cpacrMask;
newVal |= old_val & ~cpacrMask;
DPRINTF(MiscRegs, "Writing misc reg %s: %#x\n",
miscRegName[misc_reg], newVal);
}
break;
case MISCREG_CPACR_EL1:
{
const uint32_t ones = (uint32_t)(-1);
CPACR cpacrMask = 0;
cpacrMask.tta = ones;
cpacrMask.fpen = ones;
newVal &= cpacrMask;
DPRINTF(MiscRegs, "Writing misc reg %s: %#x\n",
miscRegName[misc_reg], newVal);
}
break;
case MISCREG_CPTR_EL2:
{
const uint32_t ones = (uint32_t)(-1);
CPTR cptrMask = 0;
cptrMask.tcpac = ones;
cptrMask.tta = ones;
cptrMask.tfp = ones;
newVal &= cptrMask;
cptrMask = 0;
cptrMask.res1_13_12_el2 = ones;
cptrMask.res1_9_0_el2 = ones;
newVal |= cptrMask;
DPRINTF(MiscRegs, "Writing misc reg %s: %#x\n",
miscRegName[misc_reg], newVal);
}
break;
case MISCREG_CPTR_EL3:
{
const uint32_t ones = (uint32_t)(-1);
CPTR cptrMask = 0;
cptrMask.tcpac = ones;
cptrMask.tta = ones;
cptrMask.tfp = ones;
newVal &= cptrMask;
DPRINTF(MiscRegs, "Writing misc reg %s: %#x\n",
miscRegName[misc_reg], newVal);
}
break;
case MISCREG_CSSELR:
warn_once("The csselr register isn't implemented.\n");
return;
case MISCREG_DC_ZVA_Xt:
warn("Calling DC ZVA! Not Implemeted! Expect WEIRD results\n");
return;
case MISCREG_FPSCR:
{
const uint32_t ones = (uint32_t)(-1);
FPSCR fpscrMask = 0;
fpscrMask.ioc = ones;
fpscrMask.dzc = ones;
fpscrMask.ofc = ones;
fpscrMask.ufc = ones;
fpscrMask.ixc = ones;
fpscrMask.idc = ones;
fpscrMask.ioe = ones;
fpscrMask.dze = ones;
fpscrMask.ofe = ones;
fpscrMask.ufe = ones;
fpscrMask.ixe = ones;
fpscrMask.ide = ones;
fpscrMask.len = ones;
fpscrMask.stride = ones;
fpscrMask.rMode = ones;
fpscrMask.fz = ones;
fpscrMask.dn = ones;
fpscrMask.ahp = ones;
fpscrMask.qc = ones;
fpscrMask.v = ones;
fpscrMask.c = ones;
fpscrMask.z = ones;
fpscrMask.n = ones;
newVal = (newVal & (uint32_t)fpscrMask) |
(readMiscRegNoEffect(MISCREG_FPSCR) &
~(uint32_t)fpscrMask);
tc->getDecoderPtr()->setContext(newVal);
}
break;
case MISCREG_FPSR:
{
const uint32_t ones = (uint32_t)(-1);
FPSCR fpscrMask = 0;
fpscrMask.ioc = ones;
fpscrMask.dzc = ones;
fpscrMask.ofc = ones;
fpscrMask.ufc = ones;
fpscrMask.ixc = ones;
fpscrMask.idc = ones;
fpscrMask.qc = ones;
fpscrMask.v = ones;
fpscrMask.c = ones;
fpscrMask.z = ones;
fpscrMask.n = ones;
newVal = (newVal & (uint32_t)fpscrMask) |
(readMiscRegNoEffect(MISCREG_FPSCR) &
~(uint32_t)fpscrMask);
misc_reg = MISCREG_FPSCR;
}
break;
case MISCREG_FPCR:
{
const uint32_t ones = (uint32_t)(-1);
FPSCR fpscrMask = 0;
fpscrMask.ioe = ones;
fpscrMask.dze = ones;
fpscrMask.ofe = ones;
fpscrMask.ufe = ones;
fpscrMask.ixe = ones;
fpscrMask.ide = ones;
fpscrMask.len = ones;
fpscrMask.stride = ones;
fpscrMask.rMode = ones;
fpscrMask.fz = ones;
fpscrMask.dn = ones;
fpscrMask.ahp = ones;
newVal = (newVal & (uint32_t)fpscrMask) |
(readMiscRegNoEffect(MISCREG_FPSCR) &
~(uint32_t)fpscrMask);
misc_reg = MISCREG_FPSCR;
}
break;
case MISCREG_CPSR_Q:
{
assert(!(newVal & ~CpsrMaskQ));
newVal = readMiscRegNoEffect(MISCREG_CPSR) | newVal;
misc_reg = MISCREG_CPSR;
}
break;
case MISCREG_FPSCR_QC:
{
newVal = readMiscRegNoEffect(MISCREG_FPSCR) |
(newVal & FpscrQcMask);
misc_reg = MISCREG_FPSCR;
}
break;
case MISCREG_FPSCR_EXC:
{
newVal = readMiscRegNoEffect(MISCREG_FPSCR) |
(newVal & FpscrExcMask);
misc_reg = MISCREG_FPSCR;
}
break;
case MISCREG_FPEXC:
{
// vfpv3 architecture, section B.6.1 of DDI04068
// bit 29 - valid only if fpexc[31] is 0
const uint32_t fpexcMask = 0x60000000;
newVal = (newVal & fpexcMask) |
(readMiscRegNoEffect(MISCREG_FPEXC) & ~fpexcMask);
}
break;
case MISCREG_HCR:
{
if (!haveVirtualization)
return;
}
break;
case MISCREG_IFSR:
{
// ARM ARM (ARM DDI 0406C.b) B4.1.96
const uint32_t ifsrMask =
mask(31, 13) | mask(11, 11) | mask(8, 6);
newVal = newVal & ~ifsrMask;
}
break;
case MISCREG_DFSR:
{
// ARM ARM (ARM DDI 0406C.b) B4.1.52
const uint32_t dfsrMask = mask(31, 14) | mask(8, 8);
newVal = newVal & ~dfsrMask;
}
break;
case MISCREG_AMAIR0:
case MISCREG_AMAIR1:
{
// ARM ARM (ARM DDI 0406C.b) B4.1.5
// Valid only with LPAE
if (!haveLPAE)
return;
DPRINTF(MiscRegs, "Writing AMAIR: %#x\n", newVal);
}
break;
case MISCREG_SCR:
tc->getITBPtr()->invalidateMiscReg();
tc->getDTBPtr()->invalidateMiscReg();
break;
case MISCREG_SCTLR:
{
DPRINTF(MiscRegs, "Writing SCTLR: %#x\n", newVal);
scr = readMiscRegNoEffect(MISCREG_SCR);
MiscRegIndex sctlr_idx = (haveSecurity && !scr.ns)
? MISCREG_SCTLR_S : MISCREG_SCTLR_NS;
SCTLR sctlr = miscRegs[sctlr_idx];
SCTLR new_sctlr = newVal;
new_sctlr.nmfi = ((bool)sctlr.nmfi) && !haveVirtualization;
miscRegs[sctlr_idx] = (MiscReg)new_sctlr;
tc->getITBPtr()->invalidateMiscReg();
tc->getDTBPtr()->invalidateMiscReg();
}
case MISCREG_MIDR:
case MISCREG_ID_PFR0:
case MISCREG_ID_PFR1:
case MISCREG_ID_DFR0:
case MISCREG_ID_MMFR0:
case MISCREG_ID_MMFR1:
case MISCREG_ID_MMFR2:
case MISCREG_ID_MMFR3:
case MISCREG_ID_ISAR0:
case MISCREG_ID_ISAR1:
case MISCREG_ID_ISAR2:
case MISCREG_ID_ISAR3:
case MISCREG_ID_ISAR4:
case MISCREG_ID_ISAR5:
case MISCREG_MPIDR:
case MISCREG_FPSID:
case MISCREG_TLBTR:
case MISCREG_MVFR0:
case MISCREG_MVFR1:
case MISCREG_ID_AA64AFR0_EL1:
case MISCREG_ID_AA64AFR1_EL1:
case MISCREG_ID_AA64DFR0_EL1:
case MISCREG_ID_AA64DFR1_EL1:
case MISCREG_ID_AA64ISAR0_EL1:
case MISCREG_ID_AA64ISAR1_EL1:
case MISCREG_ID_AA64MMFR0_EL1:
case MISCREG_ID_AA64MMFR1_EL1:
case MISCREG_ID_AA64PFR0_EL1:
case MISCREG_ID_AA64PFR1_EL1:
// ID registers are constants.
return;
// TLBI all entries, EL0&1 inner sharable (ignored)
case MISCREG_TLBIALLIS:
case MISCREG_TLBIALL: // TLBI all entries, EL0&1,
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
sys = tc->getSystemPtr();
for (x = 0; x < sys->numContexts(); x++) {
oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
oc->getITBPtr()->flushAllSecurity(secure_lookup, target_el);
oc->getDTBPtr()->flushAllSecurity(secure_lookup, target_el);
// If CheckerCPU is connected, need to notify it of a flush
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushAllSecurity(secure_lookup,
target_el);
checker->getDTBPtr()->flushAllSecurity(secure_lookup,
target_el);
}
}
return;
// TLBI all entries, EL0&1, instruction side
case MISCREG_ITLBIALL:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tc->getITBPtr()->flushAllSecurity(secure_lookup, target_el);
return;
// TLBI all entries, EL0&1, data side
case MISCREG_DTLBIALL:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tc->getDTBPtr()->flushAllSecurity(secure_lookup, target_el);
return;
// TLBI based on VA, EL0&1 inner sharable (ignored)
case MISCREG_TLBIMVAIS:
case MISCREG_TLBIMVA:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
sys = tc->getSystemPtr();
for (x = 0; x < sys->numContexts(); x++) {
oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
oc->getITBPtr()->flushMvaAsid(mbits(newVal, 31, 12),
bits(newVal, 7,0),
secure_lookup, target_el);
oc->getDTBPtr()->flushMvaAsid(mbits(newVal, 31, 12),
bits(newVal, 7,0),
secure_lookup, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushMvaAsid(mbits(newVal, 31, 12),
bits(newVal, 7,0), secure_lookup, target_el);
checker->getDTBPtr()->flushMvaAsid(mbits(newVal, 31, 12),
bits(newVal, 7,0), secure_lookup, target_el);
}
}
return;
// TLBI by ASID, EL0&1, inner sharable
case MISCREG_TLBIASIDIS:
case MISCREG_TLBIASID:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
sys = tc->getSystemPtr();
for (x = 0; x < sys->numContexts(); x++) {
oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
oc->getITBPtr()->flushAsid(bits(newVal, 7,0),
secure_lookup, target_el);
oc->getDTBPtr()->flushAsid(bits(newVal, 7,0),
secure_lookup, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushAsid(bits(newVal, 7,0),
secure_lookup, target_el);
checker->getDTBPtr()->flushAsid(bits(newVal, 7,0),
secure_lookup, target_el);
}
}
return;
// TLBI by address, EL0&1, inner sharable (ignored)
case MISCREG_TLBIMVAAIS:
case MISCREG_TLBIMVAA:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
hyp = 0;
tlbiMVA(tc, newVal, secure_lookup, hyp, target_el);
return;
// TLBI by address, EL2, hypervisor mode
case MISCREG_TLBIMVAH:
case MISCREG_TLBIMVAHIS:
assert32(tc);
target_el = 1; // aarch32, use hyp bit
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
hyp = 1;
tlbiMVA(tc, newVal, secure_lookup, hyp, target_el);
return;
// TLBI by address and asid, EL0&1, instruction side only
case MISCREG_ITLBIMVA:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tc->getITBPtr()->flushMvaAsid(mbits(newVal, 31, 12),
bits(newVal, 7,0), secure_lookup, target_el);
return;
// TLBI by address and asid, EL0&1, data side only
case MISCREG_DTLBIMVA:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tc->getDTBPtr()->flushMvaAsid(mbits(newVal, 31, 12),
bits(newVal, 7,0), secure_lookup, target_el);
return;
// TLBI by ASID, EL0&1, instrution side only
case MISCREG_ITLBIASID:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tc->getITBPtr()->flushAsid(bits(newVal, 7,0), secure_lookup,
target_el);
return;
// TLBI by ASID EL0&1 data size only
case MISCREG_DTLBIASID:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tc->getDTBPtr()->flushAsid(bits(newVal, 7,0), secure_lookup,
target_el);
return;
// Invalidate entire Non-secure Hyp/Non-Hyp Unified TLB
case MISCREG_TLBIALLNSNH:
case MISCREG_TLBIALLNSNHIS:
assert32(tc);
target_el = 1; // el 0 and 1 are handled together
hyp = 0;
tlbiALLN(tc, hyp, target_el);
return;
// TLBI all entries, EL2, hyp,
case MISCREG_TLBIALLH:
case MISCREG_TLBIALLHIS:
assert32(tc);
target_el = 1; // aarch32, use hyp bit
hyp = 1;
tlbiALLN(tc, hyp, target_el);
return;
// AArch64 TLBI: invalidate all entries EL3
case MISCREG_TLBI_ALLE3IS:
case MISCREG_TLBI_ALLE3:
assert64(tc);
target_el = 3;
secure_lookup = true;
tlbiALL(tc, secure_lookup, target_el);
return;
// @todo: uncomment this to enable Virtualization
// case MISCREG_TLBI_ALLE2IS:
// case MISCREG_TLBI_ALLE2:
// TLBI all entries, EL0&1
case MISCREG_TLBI_ALLE1IS:
case MISCREG_TLBI_ALLE1:
// AArch64 TLBI: invalidate all entries, stage 1, current VMID
case MISCREG_TLBI_VMALLE1IS:
case MISCREG_TLBI_VMALLE1:
// AArch64 TLBI: invalidate all entries, stages 1 & 2, current VMID
case MISCREG_TLBI_VMALLS12E1IS:
case MISCREG_TLBI_VMALLS12E1:
// @todo: handle VMID and stage 2 to enable Virtualization
assert64(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tlbiALL(tc, secure_lookup, target_el);
return;
// AArch64 TLBI: invalidate by VA and ASID, stage 1, current VMID
// VAEx(IS) and VALEx(IS) are the same because TLBs only store entries
// from the last level of translation table walks
// @todo: handle VMID to enable Virtualization
// TLBI all entries, EL0&1
case MISCREG_TLBI_VAE3IS_Xt:
case MISCREG_TLBI_VAE3_Xt:
// TLBI by VA, EL3 regime stage 1, last level walk
case MISCREG_TLBI_VALE3IS_Xt:
case MISCREG_TLBI_VALE3_Xt:
assert64(tc);
target_el = 3;
asid = 0xbeef; // does not matter, tlbi is global
secure_lookup = true;
tlbiVA(tc, newVal, asid, secure_lookup, target_el);
return;
// TLBI by VA, EL2
case MISCREG_TLBI_VAE2IS_Xt:
case MISCREG_TLBI_VAE2_Xt:
// TLBI by VA, EL2, stage1 last level walk
case MISCREG_TLBI_VALE2IS_Xt:
case MISCREG_TLBI_VALE2_Xt:
assert64(tc);
target_el = 2;
asid = 0xbeef; // does not matter, tlbi is global
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tlbiVA(tc, newVal, asid, secure_lookup, target_el);
return;
// TLBI by VA EL1 & 0, stage1, ASID, current VMID
case MISCREG_TLBI_VAE1IS_Xt:
case MISCREG_TLBI_VAE1_Xt:
case MISCREG_TLBI_VALE1IS_Xt:
case MISCREG_TLBI_VALE1_Xt:
assert64(tc);
asid = bits(newVal, 63, 48);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
tlbiVA(tc, newVal, asid, secure_lookup, target_el);
return;
// AArch64 TLBI: invalidate by ASID, stage 1, current VMID
// @todo: handle VMID to enable Virtualization
case MISCREG_TLBI_ASIDE1IS_Xt:
case MISCREG_TLBI_ASIDE1_Xt:
assert64(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
sys = tc->getSystemPtr();
for (x = 0; x < sys->numContexts(); x++) {
oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
asid = bits(newVal, 63, 48);
if (!haveLargeAsid64)
asid &= mask(8);
oc->getITBPtr()->flushAsid(asid, secure_lookup, target_el);
oc->getDTBPtr()->flushAsid(asid, secure_lookup, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushAsid(asid,
secure_lookup, target_el);
checker->getDTBPtr()->flushAsid(asid,
secure_lookup, target_el);
}
}
return;
// AArch64 TLBI: invalidate by VA, ASID, stage 1, current VMID
// VAAE1(IS) and VAALE1(IS) are the same because TLBs only store
// entries from the last level of translation table walks
// @todo: handle VMID to enable Virtualization
case MISCREG_TLBI_VAAE1IS_Xt:
case MISCREG_TLBI_VAAE1_Xt:
case MISCREG_TLBI_VAALE1IS_Xt:
case MISCREG_TLBI_VAALE1_Xt:
assert64(tc);
target_el = 1; // el 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
sys = tc->getSystemPtr();
for (x = 0; x < sys->numContexts(); x++) {
// @todo: extra controls on TLBI broadcast?
oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
Addr va = ((Addr) bits(newVal, 43, 0)) << 12;
oc->getITBPtr()->flushMva(va,
secure_lookup, false, target_el);
oc->getDTBPtr()->flushMva(va,
secure_lookup, false, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushMva(va,
secure_lookup, false, target_el);
checker->getDTBPtr()->flushMva(va,
secure_lookup, false, target_el);
}
}
return;
// AArch64 TLBI: invalidate by IPA, stage 2, current VMID
case MISCREG_TLBI_IPAS2LE1IS_Xt:
case MISCREG_TLBI_IPAS2LE1_Xt:
case MISCREG_TLBI_IPAS2E1IS_Xt:
case MISCREG_TLBI_IPAS2E1_Xt:
assert64(tc);
target_el = 1; // EL 0 and 1 are handled together
scr = readMiscReg(MISCREG_SCR, tc);
secure_lookup = haveSecurity && !scr.ns;
sys = tc->getSystemPtr();
for (x = 0; x < sys->numContexts(); x++) {
oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
Addr ipa = ((Addr) bits(newVal, 35, 0)) << 12;
oc->getITBPtr()->flushIpaVmid(ipa,
secure_lookup, false, target_el);
oc->getDTBPtr()->flushIpaVmid(ipa,
secure_lookup, false, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushIpaVmid(ipa,
secure_lookup, false, target_el);
checker->getDTBPtr()->flushIpaVmid(ipa,
secure_lookup, false, target_el);
}
}
return;
case MISCREG_ACTLR:
warn("Not doing anything for write of miscreg ACTLR\n");
break;
case MISCREG_PMXEVTYPER_PMCCFILTR:
case MISCREG_PMINTENSET_EL1 ... MISCREG_PMOVSSET_EL0:
case MISCREG_PMEVCNTR0_EL0 ... MISCREG_PMEVTYPER5_EL0:
case MISCREG_PMCR ... MISCREG_PMOVSSET:
pmu->setMiscReg(misc_reg, newVal);
break;
case MISCREG_HSTR: // TJDBX, now redifined to be RES0
{
HSTR hstrMask = 0;
hstrMask.tjdbx = 1;
newVal &= ~((uint32_t) hstrMask);
break;
}
case MISCREG_HCPTR:
{
// If a CP bit in NSACR is 0 then the corresponding bit in
// HCPTR is RAO/WI. Same applies to NSASEDIS
secure_lookup = haveSecurity &&
inSecureState(readMiscRegNoEffect(MISCREG_SCR),
readMiscRegNoEffect(MISCREG_CPSR));
if (!secure_lookup) {
MiscReg oldValue = readMiscRegNoEffect(MISCREG_HCPTR);
MiscReg mask = (readMiscRegNoEffect(MISCREG_NSACR) ^ 0x7FFF) & 0xBFFF;
newVal = (newVal & ~mask) | (oldValue & mask);
}
break;
}
case MISCREG_HDFAR: // alias for secure DFAR
misc_reg = MISCREG_DFAR_S;
break;
case MISCREG_HIFAR: // alias for secure IFAR
misc_reg = MISCREG_IFAR_S;
break;
case MISCREG_ATS1CPR:
case MISCREG_ATS1CPW:
case MISCREG_ATS1CUR:
case MISCREG_ATS1CUW:
case MISCREG_ATS12NSOPR:
case MISCREG_ATS12NSOPW:
case MISCREG_ATS12NSOUR:
case MISCREG_ATS12NSOUW:
case MISCREG_ATS1HR:
case MISCREG_ATS1HW:
{
Request::Flags flags = 0;
BaseTLB::Mode mode = BaseTLB::Read;
TLB::ArmTranslationType tranType = TLB::NormalTran;
Fault fault;
switch(misc_reg) {
case MISCREG_ATS1CPR:
flags = TLB::MustBeOne;
tranType = TLB::S1CTran;
mode = BaseTLB::Read;
break;
case MISCREG_ATS1CPW:
flags = TLB::MustBeOne;
tranType = TLB::S1CTran;
mode = BaseTLB::Write;
break;
case MISCREG_ATS1CUR:
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S1CTran;
mode = BaseTLB::Read;
break;
case MISCREG_ATS1CUW:
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S1CTran;
mode = BaseTLB::Write;
break;
case MISCREG_ATS12NSOPR:
if (!haveSecurity)
panic("Security Extensions required for ATS12NSOPR");
flags = TLB::MustBeOne;
tranType = TLB::S1S2NsTran;
mode = BaseTLB::Read;
break;
case MISCREG_ATS12NSOPW:
if (!haveSecurity)
panic("Security Extensions required for ATS12NSOPW");
flags = TLB::MustBeOne;
tranType = TLB::S1S2NsTran;
mode = BaseTLB::Write;
break;
case MISCREG_ATS12NSOUR:
if (!haveSecurity)
panic("Security Extensions required for ATS12NSOUR");
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S1S2NsTran;
mode = BaseTLB::Read;
break;
case MISCREG_ATS12NSOUW:
if (!haveSecurity)
panic("Security Extensions required for ATS12NSOUW");
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S1S2NsTran;
mode = BaseTLB::Write;
break;
case MISCREG_ATS1HR: // only really useful from secure mode.
flags = TLB::MustBeOne;
tranType = TLB::HypMode;
mode = BaseTLB::Read;
break;
case MISCREG_ATS1HW:
flags = TLB::MustBeOne;
tranType = TLB::HypMode;
mode = BaseTLB::Write;
break;
}
// If we're in timing mode then doing the translation in
// functional mode then we're slightly distorting performance
// results obtained from simulations. The translation should be
// done in the same mode the core is running in. NOTE: This
// can't be an atomic translation because that causes problems
// with unexpected atomic snoop requests.
warn("Translating via MISCREG(%d) in functional mode! Fix Me!\n", misc_reg);
Request req(0, val, 0, flags, Request::funcMasterId,
tc->pcState().pc(), tc->contextId());
fault = tc->getDTBPtr()->translateFunctional(&req, tc, mode, tranType);
TTBCR ttbcr = readMiscRegNoEffect(MISCREG_TTBCR);
HCR hcr = readMiscRegNoEffect(MISCREG_HCR);
MiscReg newVal;
if (fault == NoFault) {
Addr paddr = req.getPaddr();
if (haveLPAE && (ttbcr.eae || tranType & TLB::HypMode ||
((tranType & TLB::S1S2NsTran) && hcr.vm) )) {
newVal = (paddr & mask(39, 12)) |
(tc->getDTBPtr()->getAttr());
} else {
newVal = (paddr & 0xfffff000) |
(tc->getDTBPtr()->getAttr());
}
DPRINTF(MiscRegs,
"MISCREG: Translated addr 0x%08x: PAR: 0x%08x\n",
val, newVal);
} else {
ArmFault *armFault = reinterpret_cast<ArmFault *>(fault.get());
// Set fault bit and FSR
FSR fsr = armFault->getFsr(tc);
newVal = ((fsr >> 9) & 1) << 11;
if (newVal) {
// LPAE - rearange fault status
newVal |= ((fsr >> 0) & 0x3f) << 1;
} else {
// VMSA - rearange fault status
newVal |= ((fsr >> 0) & 0xf) << 1;
newVal |= ((fsr >> 10) & 0x1) << 5;
newVal |= ((fsr >> 12) & 0x1) << 6;
}
newVal |= 0x1; // F bit
newVal |= ((armFault->iss() >> 7) & 0x1) << 8;
newVal |= armFault->isStage2() ? 0x200 : 0;
DPRINTF(MiscRegs,
"MISCREG: Translated addr 0x%08x fault fsr %#x: PAR: 0x%08x\n",
val, fsr, newVal);
}
setMiscRegNoEffect(MISCREG_PAR, newVal);
return;
}
case MISCREG_TTBCR:
{
TTBCR ttbcr = readMiscRegNoEffect(MISCREG_TTBCR);
const uint32_t ones = (uint32_t)(-1);
TTBCR ttbcrMask = 0;
TTBCR ttbcrNew = newVal;
// ARM DDI 0406C.b, ARMv7-32
ttbcrMask.n = ones; // T0SZ
if (haveSecurity) {
ttbcrMask.pd0 = ones;
ttbcrMask.pd1 = ones;
}
ttbcrMask.epd0 = ones;
ttbcrMask.irgn0 = ones;
ttbcrMask.orgn0 = ones;
ttbcrMask.sh0 = ones;
ttbcrMask.ps = ones; // T1SZ
ttbcrMask.a1 = ones;
ttbcrMask.epd1 = ones;
ttbcrMask.irgn1 = ones;
ttbcrMask.orgn1 = ones;
ttbcrMask.sh1 = ones;
if (haveLPAE)
ttbcrMask.eae = ones;
if (haveLPAE && ttbcrNew.eae) {
newVal = newVal & ttbcrMask;
} else {
newVal = (newVal & ttbcrMask) | (ttbcr & (~ttbcrMask));
}
}
case MISCREG_TTBR0:
case MISCREG_TTBR1:
{
TTBCR ttbcr = readMiscRegNoEffect(MISCREG_TTBCR);
if (haveLPAE) {
if (ttbcr.eae) {
// ARMv7 bit 63-56, 47-40 reserved, UNK/SBZP
// ARMv8 AArch32 bit 63-56 only
uint64_t ttbrMask = mask(63,56) | mask(47,40);
newVal = (newVal & (~ttbrMask));
}
}
}
case MISCREG_SCTLR_EL1:
{
tc->getITBPtr()->invalidateMiscReg();
tc->getDTBPtr()->invalidateMiscReg();
setMiscRegNoEffect(misc_reg, newVal);
}
case MISCREG_CONTEXTIDR:
case MISCREG_PRRR:
case MISCREG_NMRR:
case MISCREG_MAIR0:
case MISCREG_MAIR1:
case MISCREG_DACR:
case MISCREG_VTTBR:
case MISCREG_SCR_EL3:
case MISCREG_HCR_EL2:
case MISCREG_TCR_EL1:
case MISCREG_TCR_EL2:
case MISCREG_TCR_EL3:
case MISCREG_SCTLR_EL2:
case MISCREG_SCTLR_EL3:
case MISCREG_HSCTLR:
case MISCREG_TTBR0_EL1:
case MISCREG_TTBR1_EL1:
case MISCREG_TTBR0_EL2:
case MISCREG_TTBR0_EL3:
tc->getITBPtr()->invalidateMiscReg();
tc->getDTBPtr()->invalidateMiscReg();
break;
case MISCREG_NZCV:
{
CPSR cpsr = val;
tc->setCCReg(CCREG_NZ, cpsr.nz);
tc->setCCReg(CCREG_C, cpsr.c);
tc->setCCReg(CCREG_V, cpsr.v);
}
break;
case MISCREG_DAIF:
{
CPSR cpsr = miscRegs[MISCREG_CPSR];
cpsr.daif = (uint8_t) ((CPSR) newVal).daif;
newVal = cpsr;
misc_reg = MISCREG_CPSR;
}
break;
case MISCREG_SP_EL0:
tc->setIntReg(INTREG_SP0, newVal);
break;
case MISCREG_SP_EL1:
tc->setIntReg(INTREG_SP1, newVal);
break;
case MISCREG_SP_EL2:
tc->setIntReg(INTREG_SP2, newVal);
break;
case MISCREG_SPSEL:
{
CPSR cpsr = miscRegs[MISCREG_CPSR];
cpsr.sp = (uint8_t) ((CPSR) newVal).sp;
newVal = cpsr;
misc_reg = MISCREG_CPSR;
}
break;
case MISCREG_CURRENTEL:
{
CPSR cpsr = miscRegs[MISCREG_CPSR];
cpsr.el = (uint8_t) ((CPSR) newVal).el;
newVal = cpsr;
misc_reg = MISCREG_CPSR;
}
break;
case MISCREG_AT_S1E1R_Xt:
case MISCREG_AT_S1E1W_Xt:
case MISCREG_AT_S1E0R_Xt:
case MISCREG_AT_S1E0W_Xt:
case MISCREG_AT_S1E2R_Xt:
case MISCREG_AT_S1E2W_Xt:
case MISCREG_AT_S12E1R_Xt:
case MISCREG_AT_S12E1W_Xt:
case MISCREG_AT_S12E0R_Xt:
case MISCREG_AT_S12E0W_Xt:
case MISCREG_AT_S1E3R_Xt:
case MISCREG_AT_S1E3W_Xt:
{
RequestPtr req = new Request;
Request::Flags flags = 0;
BaseTLB::Mode mode = BaseTLB::Read;
TLB::ArmTranslationType tranType = TLB::NormalTran;
Fault fault;
switch(misc_reg) {
case MISCREG_AT_S1E1R_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S1E1Tran;
mode = BaseTLB::Read;
break;
case MISCREG_AT_S1E1W_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S1E1Tran;
mode = BaseTLB::Write;
break;
case MISCREG_AT_S1E0R_Xt:
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S1E0Tran;
mode = BaseTLB::Read;
break;
case MISCREG_AT_S1E0W_Xt:
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S1E0Tran;
mode = BaseTLB::Write;
break;
case MISCREG_AT_S1E2R_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S1E2Tran;
mode = BaseTLB::Read;
break;
case MISCREG_AT_S1E2W_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S1E2Tran;
mode = BaseTLB::Write;
break;
case MISCREG_AT_S12E0R_Xt:
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S12E0Tran;
mode = BaseTLB::Read;
break;
case MISCREG_AT_S12E0W_Xt:
flags = TLB::MustBeOne | TLB::UserMode;
tranType = TLB::S12E0Tran;
mode = BaseTLB::Write;
break;
case MISCREG_AT_S12E1R_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S12E1Tran;
mode = BaseTLB::Read;
break;
case MISCREG_AT_S12E1W_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S12E1Tran;
mode = BaseTLB::Write;
break;
case MISCREG_AT_S1E3R_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S1E3Tran;
mode = BaseTLB::Read;
break;
case MISCREG_AT_S1E3W_Xt:
flags = TLB::MustBeOne;
tranType = TLB::S1E3Tran;
mode = BaseTLB::Write;
break;
}
// If we're in timing mode then doing the translation in
// functional mode then we're slightly distorting performance
// results obtained from simulations. The translation should be
// done in the same mode the core is running in. NOTE: This
// can't be an atomic translation because that causes problems
// with unexpected atomic snoop requests.
warn("Translating via MISCREG(%d) in functional mode! Fix Me!\n", misc_reg);
req->setVirt(0, val, 0, flags, Request::funcMasterId,
tc->pcState().pc());
req->setContext(tc->contextId());
fault = tc->getDTBPtr()->translateFunctional(req, tc, mode,
tranType);
MiscReg newVal;
if (fault == NoFault) {
Addr paddr = req->getPaddr();
uint64_t attr = tc->getDTBPtr()->getAttr();
uint64_t attr1 = attr >> 56;
if (!attr1 || attr1 ==0x44) {
attr |= 0x100;
attr &= ~ uint64_t(0x80);
}
newVal = (paddr & mask(47, 12)) | attr;
DPRINTF(MiscRegs,
"MISCREG: Translated addr %#x: PAR_EL1: %#xx\n",
val, newVal);
} else {
ArmFault *armFault = reinterpret_cast<ArmFault *>(fault.get());
// Set fault bit and FSR
FSR fsr = armFault->getFsr(tc);
CPSR cpsr = tc->readMiscReg(MISCREG_CPSR);
if (cpsr.width) { // AArch32
newVal = ((fsr >> 9) & 1) << 11;
// rearrange fault status
newVal |= ((fsr >> 0) & 0x3f) << 1;
newVal |= 0x1; // F bit
newVal |= ((armFault->iss() >> 7) & 0x1) << 8;
newVal |= armFault->isStage2() ? 0x200 : 0;
} else { // AArch64
newVal = 1; // F bit
newVal |= fsr << 1; // FST
// TODO: DDI 0487A.f D7-2083, AbortFault's s1ptw bit.
newVal |= armFault->isStage2() ? 1 << 8 : 0; // PTW
newVal |= armFault->isStage2() ? 1 << 9 : 0; // S
newVal |= 1 << 11; // RES1
}
DPRINTF(MiscRegs,
"MISCREG: Translated addr %#x fault fsr %#x: PAR: %#x\n",
val, fsr, newVal);
}
delete req;
setMiscRegNoEffect(MISCREG_PAR_EL1, newVal);
return;
}
case MISCREG_SPSR_EL3:
case MISCREG_SPSR_EL2:
case MISCREG_SPSR_EL1:
// Force bits 23:21 to 0
newVal = val & ~(0x7 << 21);
break;
case MISCREG_L2CTLR:
warn("miscreg L2CTLR (%s) written with %#x. ignored...\n",
miscRegName[misc_reg], uint32_t(val));
break;
// Generic Timer registers
case MISCREG_CNTFRQ ... MISCREG_CNTHP_CTL:
case MISCREG_CNTPCT ... MISCREG_CNTHP_CVAL:
case MISCREG_CNTKCTL_EL1 ... MISCREG_CNTV_CVAL_EL0:
case MISCREG_CNTVOFF_EL2 ... MISCREG_CNTPS_CVAL_EL1:
getGenericTimer(tc).setMiscReg(misc_reg, newVal);
break;
}
}
setMiscRegNoEffect(misc_reg, newVal);
}
void
ISA::tlbiVA(ThreadContext *tc, MiscReg newVal, uint16_t asid,
bool secure_lookup, uint8_t target_el)
{
if (!haveLargeAsid64)
asid &= mask(8);
Addr va = ((Addr) bits(newVal, 43, 0)) << 12;
System *sys = tc->getSystemPtr();
for (int x = 0; x < sys->numContexts(); x++) {
ThreadContext *oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
oc->getITBPtr()->flushMvaAsid(va, asid,
secure_lookup, target_el);
oc->getDTBPtr()->flushMvaAsid(va, asid,
secure_lookup, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushMvaAsid(
va, asid, secure_lookup, target_el);
checker->getDTBPtr()->flushMvaAsid(
va, asid, secure_lookup, target_el);
}
}
}
void
ISA::tlbiALL(ThreadContext *tc, bool secure_lookup, uint8_t target_el)
{
System *sys = tc->getSystemPtr();
for (int x = 0; x < sys->numContexts(); x++) {
ThreadContext *oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
oc->getITBPtr()->flushAllSecurity(secure_lookup, target_el);
oc->getDTBPtr()->flushAllSecurity(secure_lookup, target_el);
// If CheckerCPU is connected, need to notify it of a flush
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushAllSecurity(secure_lookup,
target_el);
checker->getDTBPtr()->flushAllSecurity(secure_lookup,
target_el);
}
}
}
void
ISA::tlbiALLN(ThreadContext *tc, bool hyp, uint8_t target_el)
{
System *sys = tc->getSystemPtr();
for (int x = 0; x < sys->numContexts(); x++) {
ThreadContext *oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
oc->getITBPtr()->flushAllNs(hyp, target_el);
oc->getDTBPtr()->flushAllNs(hyp, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushAllNs(hyp, target_el);
checker->getDTBPtr()->flushAllNs(hyp, target_el);
}
}
}
void
ISA::tlbiMVA(ThreadContext *tc, MiscReg newVal, bool secure_lookup, bool hyp,
uint8_t target_el)
{
System *sys = tc->getSystemPtr();
for (int x = 0; x < sys->numContexts(); x++) {
ThreadContext *oc = sys->getThreadContext(x);
assert(oc->getITBPtr() && oc->getDTBPtr());
oc->getITBPtr()->flushMva(mbits(newVal, 31,12),
secure_lookup, hyp, target_el);
oc->getDTBPtr()->flushMva(mbits(newVal, 31,12),
secure_lookup, hyp, target_el);
CheckerCPU *checker = oc->getCheckerCpuPtr();
if (checker) {
checker->getITBPtr()->flushMva(mbits(newVal, 31,12),
secure_lookup, hyp, target_el);
checker->getDTBPtr()->flushMva(mbits(newVal, 31,12),
secure_lookup, hyp, target_el);
}
}
}
BaseISADevice &
ISA::getGenericTimer(ThreadContext *tc)
{
// We only need to create an ISA interface the first time we try
// to access the timer.
if (timer)
return *timer.get();
assert(system);
GenericTimer *generic_timer(system->getGenericTimer());
if (!generic_timer) {
panic("Trying to get a generic timer from a system that hasn't "
"been configured to use a generic timer.\n");
}
timer.reset(new GenericTimerISA(*generic_timer, tc->contextId()));
return *timer.get();
}
}
ArmISA::ISA *
ArmISAParams::create()
{
return new ArmISA::ISA(this);
}