sanchayanmaity/gem5
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
gem5/arch/alpha/isa/decoder.isa
Steve Reinhardt 3923eec0ef Change how memory operands are handled in ISA descriptions. 
Should enable implementation of split-phase timing loads
with new memory model.
May create slight timing differences under FullCPU, as I
believe we were not handling software prefetches correctly
before when the split MemAcc/Exec model was used.  I haven't
looked into this in any detail though.

arch/alpha/isa/decoder.isa:
    HwLoadStore format split into separate HwLoad and
    HwStore formats.
    Copy instructions now fall under MiscPrefetch format.
    Mem_write_result is now just write_result in store
    conditionals.
arch/alpha/isa/mem.isa:
    Split MemAccExecute and LoadStoreExecute templates
    into separate templates for loads and stores; now
    that memory operands are handled differently from
    registers, it's impossible to have a single template
    serve both.
    Also unified the handling of "regular" prefetches
    (loads to r31) and "misc" prefetches (e.g., wh64)
    under the new scheme.  It looks like SW prefetches
    were not handled correctly in FullCPU up til now,
    since we generated an execute() method for the outer
    instruction but didn't generate a proper method for
    MemAcc::execute() (instead getting a default no-op
    method for that).
arch/alpha/isa/pal.isa:
    Split HwLoadStore into separate HwLoad and HwStore
    formats to select proper template (see change to
    mem.isa in this changeset).
arch/isa_parser.py:
    Stop trying to treat memory operands like register
    operands, since we never used them in a uniform way
    anyway, and it made it impossible to do split-phase
    loads as needed for the new CPU model.  Now there's no
    more 'op_mem_rd', 'op_nonmem_rd', etc.: 'op_rd' just does
    register operands, and the template code is responsible
    for formulating the call to the memory system.  Right now
    the only thing exported by InstObjParams is a new attribute
    'mem_acc_size' which gives the memory access size in bits,
    though more attributes can be added if needed.

    Also moved code in findOperands() method to
    OperandDescriptorList.__init__(), which is where it belongs.

--HG--
extra : convert_revision : 6d53d07e0c5e828455834ded4395fa40f9146a34
2006-02-10 09:12:55 -05:00

801 lines
32 KiB
C++
Raw Blame History

// -*- mode:c++ -*-
// Copyright (c) 2003-2005 The Regents of The University of Michigan
// All rights reserved.
//
// 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.
decode OPCODE default Unknown::unknown() {
format LoadAddress {
0x08: lda({{ Ra = Rb + disp; }});
0x09: ldah({{ Ra = Rb + (disp << 16); }});
}
format LoadOrNop {
0x0a: ldbu({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.ub; }});
0x0c: ldwu({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.uw; }});
0x0b: ldq_u({{ EA = (Rb + disp) & ~7; }}, {{ Ra = Mem.uq; }});
0x23: ldt({{ EA = Rb + disp; }}, {{ Fa = Mem.df; }});
0x2a: ldl_l({{ EA = Rb + disp; }}, {{ Ra.sl = Mem.sl; }}, LOCKED);
0x2b: ldq_l({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.uq; }}, LOCKED);
0x20: MiscPrefetch::copy_load({{ EA = Ra; }},
{{ fault = xc->copySrcTranslate(EA); }},
IsMemRef, IsLoad, IsCopy);
}
format LoadOrPrefetch {
0x28: ldl({{ EA = Rb + disp; }}, {{ Ra.sl = Mem.sl; }});
0x29: ldq({{ EA = Rb + disp; }}, {{ Ra.uq = Mem.uq; }}, EVICT_NEXT);
// IsFloating flag on lds gets the prefetch to disassemble
// using f31 instead of r31... funcitonally it's unnecessary
0x22: lds({{ EA = Rb + disp; }}, {{ Fa.uq = s_to_t(Mem.ul); }},
PF_EXCLUSIVE, IsFloating);
}
format Store {
0x0e: stb({{ EA = Rb + disp; }}, {{ Mem.ub = Ra<7:0>; }});
0x0d: stw({{ EA = Rb + disp; }}, {{ Mem.uw = Ra<15:0>; }});
0x2c: stl({{ EA = Rb + disp; }}, {{ Mem.ul = Ra<31:0>; }});
0x2d: stq({{ EA = Rb + disp; }}, {{ Mem.uq = Ra.uq; }});
0x0f: stq_u({{ EA = (Rb + disp) & ~7; }}, {{ Mem.uq = Ra.uq; }});
0x26: sts({{ EA = Rb + disp; }}, {{ Mem.ul = t_to_s(Fa.uq); }});
0x27: stt({{ EA = Rb + disp; }}, {{ Mem.df = Fa; }});
0x24: MiscPrefetch::copy_store({{ EA = Rb; }},
{{ fault = xc->copy(EA); }},
IsMemRef, IsStore, IsCopy);
}
format StoreCond {
0x2e: stl_c({{ EA = Rb + disp; }}, {{ Mem.ul = Ra<31:0>; }},
{{
uint64_t tmp = write_result;
// see stq_c
Ra = (tmp == 0 || tmp == 1) ? tmp : Ra;
}}, LOCKED);
0x2f: stq_c({{ EA = Rb + disp; }}, {{ Mem.uq = Ra; }},
{{
uint64_t tmp = write_result;
// If the write operation returns 0 or 1, then
// this was a conventional store conditional,
// and the value indicates the success/failure
// of the operation. If another value is
// returned, then this was a Turbolaser
// mailbox access, and we don't update the
// result register at all.
Ra = (tmp == 0 || tmp == 1) ? tmp : Ra;
}}, LOCKED);
}
format IntegerOperate {
0x10: decode INTFUNC { // integer arithmetic operations
0x00: addl({{ Rc.sl = Ra.sl + Rb_or_imm.sl; }});
0x40: addlv({{
uint32_t tmp = Ra.sl + Rb_or_imm.sl;
// signed overflow occurs when operands have same sign
// and sign of result does not match.
if (Ra.sl<31:> == Rb_or_imm.sl<31:> && tmp<31:> != Ra.sl<31:>)
fault = Integer_Overflow_Fault;
Rc.sl = tmp;
}});
0x02: s4addl({{ Rc.sl = (Ra.sl << 2) + Rb_or_imm.sl; }});
0x12: s8addl({{ Rc.sl = (Ra.sl << 3) + Rb_or_imm.sl; }});
0x20: addq({{ Rc = Ra + Rb_or_imm; }});
0x60: addqv({{
uint64_t tmp = Ra + Rb_or_imm;
// signed overflow occurs when operands have same sign
// and sign of result does not match.
if (Ra<63:> == Rb_or_imm<63:> && tmp<63:> != Ra<63:>)
fault = Integer_Overflow_Fault;
Rc = tmp;
}});
0x22: s4addq({{ Rc = (Ra << 2) + Rb_or_imm; }});
0x32: s8addq({{ Rc = (Ra << 3) + Rb_or_imm; }});
0x09: subl({{ Rc.sl = Ra.sl - Rb_or_imm.sl; }});
0x49: sublv({{
uint32_t tmp = Ra.sl - Rb_or_imm.sl;
// signed overflow detection is same as for add,
// except we need to look at the *complemented*
// sign bit of the subtrahend (Rb), i.e., if the initial
// signs are the *same* then no overflow can occur
if (Ra.sl<31:> != Rb_or_imm.sl<31:> && tmp<31:> != Ra.sl<31:>)
fault = Integer_Overflow_Fault;
Rc.sl = tmp;
}});
0x0b: s4subl({{ Rc.sl = (Ra.sl << 2) - Rb_or_imm.sl; }});
0x1b: s8subl({{ Rc.sl = (Ra.sl << 3) - Rb_or_imm.sl; }});
0x29: subq({{ Rc = Ra - Rb_or_imm; }});
0x69: subqv({{
uint64_t tmp = Ra - Rb_or_imm;
// signed overflow detection is same as for add,
// except we need to look at the *complemented*
// sign bit of the subtrahend (Rb), i.e., if the initial
// signs are the *same* then no overflow can occur
if (Ra<63:> != Rb_or_imm<63:> && tmp<63:> != Ra<63:>)
fault = Integer_Overflow_Fault;
Rc = tmp;
}});
0x2b: s4subq({{ Rc = (Ra << 2) - Rb_or_imm; }});
0x3b: s8subq({{ Rc = (Ra << 3) - Rb_or_imm; }});
0x2d: cmpeq({{ Rc = (Ra == Rb_or_imm); }});
0x6d: cmple({{ Rc = (Ra.sq <= Rb_or_imm.sq); }});
0x4d: cmplt({{ Rc = (Ra.sq < Rb_or_imm.sq); }});
0x3d: cmpule({{ Rc = (Ra.uq <= Rb_or_imm.uq); }});
0x1d: cmpult({{ Rc = (Ra.uq < Rb_or_imm.uq); }});
0x0f: cmpbge({{
int hi = 7;
int lo = 0;
uint64_t tmp = 0;
for (int i = 0; i < 8; ++i) {
tmp |= (Ra.uq<hi:lo> >= Rb_or_imm.uq<hi:lo>) << i;
hi += 8;
lo += 8;
}
Rc = tmp;
}});
}
0x11: decode INTFUNC { // integer logical operations
0x00: and({{ Rc = Ra & Rb_or_imm; }});
0x08: bic({{ Rc = Ra & ~Rb_or_imm; }});
0x20: bis({{ Rc = Ra | Rb_or_imm; }});
0x28: ornot({{ Rc = Ra | ~Rb_or_imm; }});
0x40: xor({{ Rc = Ra ^ Rb_or_imm; }});
0x48: eqv({{ Rc = Ra ^ ~Rb_or_imm; }});
// conditional moves
0x14: cmovlbs({{ Rc = ((Ra & 1) == 1) ? Rb_or_imm : Rc; }});
0x16: cmovlbc({{ Rc = ((Ra & 1) == 0) ? Rb_or_imm : Rc; }});
0x24: cmoveq({{ Rc = (Ra == 0) ? Rb_or_imm : Rc; }});
0x26: cmovne({{ Rc = (Ra != 0) ? Rb_or_imm : Rc; }});
0x44: cmovlt({{ Rc = (Ra.sq < 0) ? Rb_or_imm : Rc; }});
0x46: cmovge({{ Rc = (Ra.sq >= 0) ? Rb_or_imm : Rc; }});
0x64: cmovle({{ Rc = (Ra.sq <= 0) ? Rb_or_imm : Rc; }});
0x66: cmovgt({{ Rc = (Ra.sq > 0) ? Rb_or_imm : Rc; }});
// For AMASK, RA must be R31.
0x61: decode RA {
31: amask({{ Rc = Rb_or_imm & ~ULL(0x17); }});
}
// For IMPLVER, RA must be R31 and the B operand
// must be the immediate value 1.
0x6c: decode RA {
31: decode IMM {
1: decode INTIMM {
// return EV5 for FULL_SYSTEM and EV6 otherwise
1: implver({{
#if FULL_SYSTEM
Rc = 1;
#else
Rc = 2;
#endif
}});
}
}
}
#if FULL_SYSTEM
// The mysterious 11.25...
0x25: WarnUnimpl::eleven25();
#endif
}
0x12: decode INTFUNC {
0x39: sll({{ Rc = Ra << Rb_or_imm<5:0>; }});
0x34: srl({{ Rc = Ra.uq >> Rb_or_imm<5:0>; }});
0x3c: sra({{ Rc = Ra.sq >> Rb_or_imm<5:0>; }});
0x02: mskbl({{ Rc = Ra & ~(mask( 8) << (Rb_or_imm<2:0> * 8)); }});
0x12: mskwl({{ Rc = Ra & ~(mask(16) << (Rb_or_imm<2:0> * 8)); }});
0x22: mskll({{ Rc = Ra & ~(mask(32) << (Rb_or_imm<2:0> * 8)); }});
0x32: mskql({{ Rc = Ra & ~(mask(64) << (Rb_or_imm<2:0> * 8)); }});
0x52: mskwh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra & ~(mask(16) >> (64 - 8 * bv))) : Ra;
}});
0x62: msklh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra & ~(mask(32) >> (64 - 8 * bv))) : Ra;
}});
0x72: mskqh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra & ~(mask(64) >> (64 - 8 * bv))) : Ra;
}});
0x06: extbl({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8))< 7:0>; }});
0x16: extwl({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8))<15:0>; }});
0x26: extll({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8))<31:0>; }});
0x36: extql({{ Rc = (Ra.uq >> (Rb_or_imm<2:0> * 8)); }});
0x5a: extwh({{
Rc = (Ra << (64 - (Rb_or_imm<2:0> * 8))<5:0>)<15:0>; }});
0x6a: extlh({{
Rc = (Ra << (64 - (Rb_or_imm<2:0> * 8))<5:0>)<31:0>; }});
0x7a: extqh({{
Rc = (Ra << (64 - (Rb_or_imm<2:0> * 8))<5:0>); }});
0x0b: insbl({{ Rc = Ra< 7:0> << (Rb_or_imm<2:0> * 8); }});
0x1b: inswl({{ Rc = Ra<15:0> << (Rb_or_imm<2:0> * 8); }});
0x2b: insll({{ Rc = Ra<31:0> << (Rb_or_imm<2:0> * 8); }});
0x3b: insql({{ Rc = Ra << (Rb_or_imm<2:0> * 8); }});
0x57: inswh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra.uq<15:0> >> (64 - 8 * bv)) : 0;
}});
0x67: inslh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra.uq<31:0> >> (64 - 8 * bv)) : 0;
}});
0x77: insqh({{
int bv = Rb_or_imm<2:0>;
Rc = bv ? (Ra.uq >> (64 - 8 * bv)) : 0;
}});
0x30: zap({{
uint64_t zapmask = 0;
for (int i = 0; i < 8; ++i) {
if (Rb_or_imm<i:>)
zapmask |= (mask(8) << (i * 8));
}
Rc = Ra & ~zapmask;
}});
0x31: zapnot({{
uint64_t zapmask = 0;
for (int i = 0; i < 8; ++i) {
if (!Rb_or_imm<i:>)
zapmask |= (mask(8) << (i * 8));
}
Rc = Ra & ~zapmask;
}});
}
0x13: decode INTFUNC { // integer multiplies
0x00: mull({{ Rc.sl = Ra.sl * Rb_or_imm.sl; }}, IntMultOp);
0x20: mulq({{ Rc = Ra * Rb_or_imm; }}, IntMultOp);
0x30: umulh({{
uint64_t hi, lo;
mul128(Ra, Rb_or_imm, hi, lo);
Rc = hi;
}}, IntMultOp);
0x40: mullv({{
// 32-bit multiply with trap on overflow
int64_t Rax = Ra.sl; // sign extended version of Ra.sl
int64_t Rbx = Rb_or_imm.sl;
int64_t tmp = Rax * Rbx;
// To avoid overflow, all the upper 32 bits must match
// the sign bit of the lower 32. We code this as
// checking the upper 33 bits for all 0s or all 1s.
uint64_t sign_bits = tmp<63:31>;
if (sign_bits != 0 && sign_bits != mask(33))
fault = Integer_Overflow_Fault;
Rc.sl = tmp<31:0>;
}}, IntMultOp);
0x60: mulqv({{
// 64-bit multiply with trap on overflow
uint64_t hi, lo;
mul128(Ra, Rb_or_imm, hi, lo);
// all the upper 64 bits must match the sign bit of
// the lower 64
if (!((hi == 0 && lo<63:> == 0) ||
(hi == mask(64) && lo<63:> == 1)))
fault = Integer_Overflow_Fault;
Rc = lo;
}}, IntMultOp);
}
0x1c: decode INTFUNC {
0x00: decode RA { 31: sextb({{ Rc.sb = Rb_or_imm< 7:0>; }}); }
0x01: decode RA { 31: sextw({{ Rc.sw = Rb_or_imm<15:0>; }}); }
0x32: ctlz({{
uint64_t count = 0;
uint64_t temp = Rb;
if (temp<63:32>) temp >>= 32; else count += 32;
if (temp<31:16>) temp >>= 16; else count += 16;
if (temp<15:8>) temp >>= 8; else count += 8;
if (temp<7:4>) temp >>= 4; else count += 4;
if (temp<3:2>) temp >>= 2; else count += 2;
if (temp<1:1>) temp >>= 1; else count += 1;
if ((temp<0:0>) != 0x1) count += 1;
Rc = count;
}}, IntAluOp);
0x33: cttz({{
uint64_t count = 0;
uint64_t temp = Rb;
if (!(temp<31:0>)) { temp >>= 32; count += 32; }
if (!(temp<15:0>)) { temp >>= 16; count += 16; }
if (!(temp<7:0>)) { temp >>= 8; count += 8; }
if (!(temp<3:0>)) { temp >>= 4; count += 4; }
if (!(temp<1:0>)) { temp >>= 2; count += 2; }
if (!(temp<0:0> & ULL(0x1))) count += 1;
Rc = count;
}}, IntAluOp);
format FailUnimpl {
0x30: ctpop();
0x31: perr();
0x34: unpkbw();
0x35: unpkbl();
0x36: pkwb();
0x37: pklb();
0x38: minsb8();
0x39: minsw4();
0x3a: minub8();
0x3b: minuw4();
0x3c: maxub8();
0x3d: maxuw4();
0x3e: maxsb8();
0x3f: maxsw4();
}
format BasicOperateWithNopCheck {
0x70: decode RB {
31: ftoit({{ Rc = Fa.uq; }}, FloatCvtOp);
}
0x78: decode RB {
31: ftois({{ Rc.sl = t_to_s(Fa.uq); }},
FloatCvtOp);
}
}
}
}
// Conditional branches.
format CondBranch {
0x39: beq({{ cond = (Ra == 0); }});
0x3d: bne({{ cond = (Ra != 0); }});
0x3e: bge({{ cond = (Ra.sq >= 0); }});
0x3f: bgt({{ cond = (Ra.sq > 0); }});
0x3b: ble({{ cond = (Ra.sq <= 0); }});
0x3a: blt({{ cond = (Ra.sq < 0); }});
0x38: blbc({{ cond = ((Ra & 1) == 0); }});
0x3c: blbs({{ cond = ((Ra & 1) == 1); }});
0x31: fbeq({{ cond = (Fa == 0); }});
0x35: fbne({{ cond = (Fa != 0); }});
0x36: fbge({{ cond = (Fa >= 0); }});
0x37: fbgt({{ cond = (Fa > 0); }});
0x33: fble({{ cond = (Fa <= 0); }});
0x32: fblt({{ cond = (Fa < 0); }});
}
// unconditional branches
format UncondBranch {
0x30: br();
0x34: bsr(IsCall);
}
// indirect branches
0x1a: decode JMPFUNC {
format Jump {
0: jmp();
1: jsr(IsCall);
2: ret(IsReturn);
3: jsr_coroutine(IsCall, IsReturn);
}
}
// Square root and integer-to-FP moves
0x14: decode FP_SHORTFUNC {
// Integer to FP register moves must have RB == 31
0x4: decode RB {
31: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
0x004: itofs({{ Fc.uq = s_to_t(Ra.ul); }}, FloatCvtOp);
0x024: itoft({{ Fc.uq = Ra.uq; }}, FloatCvtOp);
0x014: FailUnimpl::itoff(); // VAX-format conversion
}
}
}
// Square root instructions must have FA == 31
0xb: decode FA {
31: decode FP_TYPEFUNC {
format FloatingPointOperate {
#if SS_COMPATIBLE_FP
0x0b: sqrts({{
if (Fb < 0.0)
fault = Arithmetic_Fault;
Fc = sqrt(Fb);
}}, FloatSqrtOp);
#else
0x0b: sqrts({{
if (Fb.sf < 0.0)
fault = Arithmetic_Fault;
Fc.sf = sqrt(Fb.sf);
}}, FloatSqrtOp);
#endif
0x2b: sqrtt({{
if (Fb < 0.0)
fault = Arithmetic_Fault;
Fc = sqrt(Fb);
}}, FloatSqrtOp);
}
}
}
// VAX-format sqrtf and sqrtg are not implemented
0xa: FailUnimpl::sqrtfg();
}
// IEEE floating point
0x16: decode FP_SHORTFUNC_TOP2 {
// The top two bits of the short function code break this
// space into four groups: binary ops, compares, reserved, and
// conversions. See Table 4-12 of AHB. There are different
// special cases in these different groups, so we decode on
// these top two bits first just to select a decode strategy.
// Most of these instructions may have various trapping and
// rounding mode flags set; these are decoded in the
// FloatingPointDecode template used by the
// FloatingPointOperate format.
// add/sub/mul/div: just decode on the short function code
// and source type. All valid trapping and rounding modes apply.
0: decode FP_TRAPMODE {
// check for valid trapping modes here
0,1,5,7: decode FP_TYPEFUNC {
format FloatingPointOperate {
#if SS_COMPATIBLE_FP
0x00: adds({{ Fc = Fa + Fb; }});
0x01: subs({{ Fc = Fa - Fb; }});
0x02: muls({{ Fc = Fa * Fb; }}, FloatMultOp);
0x03: divs({{ Fc = Fa / Fb; }}, FloatDivOp);
#else
0x00: adds({{ Fc.sf = Fa.sf + Fb.sf; }});
0x01: subs({{ Fc.sf = Fa.sf - Fb.sf; }});
0x02: muls({{ Fc.sf = Fa.sf * Fb.sf; }}, FloatMultOp);
0x03: divs({{ Fc.sf = Fa.sf / Fb.sf; }}, FloatDivOp);
#endif
0x20: addt({{ Fc = Fa + Fb; }});
0x21: subt({{ Fc = Fa - Fb; }});
0x22: mult({{ Fc = Fa * Fb; }}, FloatMultOp);
0x23: divt({{ Fc = Fa / Fb; }}, FloatDivOp);
}
}
}
// Floating-point compare instructions must have the default
// rounding mode, and may use the default trapping mode or
// /SU. Both trapping modes are treated the same by M5; the
// only difference on the real hardware (as far a I can tell)
// is that without /SU you'd get an imprecise trap if you
// tried to compare a NaN with something else (instead of an
// "unordered" result).
1: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
0x0a5, 0x5a5: cmpteq({{ Fc = (Fa == Fb) ? 2.0 : 0.0; }},
FloatCmpOp);
0x0a7, 0x5a7: cmptle({{ Fc = (Fa <= Fb) ? 2.0 : 0.0; }},
FloatCmpOp);
0x0a6, 0x5a6: cmptlt({{ Fc = (Fa < Fb) ? 2.0 : 0.0; }},
FloatCmpOp);
0x0a4, 0x5a4: cmptun({{ // unordered
Fc = (!(Fa < Fb) && !(Fa == Fb) && !(Fa > Fb)) ? 2.0 : 0.0;
}}, FloatCmpOp);
}
}
// The FP-to-integer and integer-to-FP conversion insts
// require that FA be 31.
3: decode FA {
31: decode FP_TYPEFUNC {
format FloatingPointOperate {
0x2f: decode FP_ROUNDMODE {
format FPFixedRounding {
// "chopped" i.e. round toward zero
0: cvttq({{ Fc.sq = (int64_t)trunc(Fb); }},
Chopped);
// round to minus infinity
1: cvttq({{ Fc.sq = (int64_t)floor(Fb); }},
MinusInfinity);
}
default: cvttq({{ Fc.sq = (int64_t)nearbyint(Fb); }});
}
// The cvtts opcode is overloaded to be cvtst if the trap
// mode is 2 or 6 (which are not valid otherwise)
0x2c: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
// trap on denorm version "cvtst/s" is
// simulated same as cvtst
0x2ac, 0x6ac: cvtst({{ Fc = Fb.sf; }});
}
default: cvtts({{ Fc.sf = Fb; }});
}
// The trapping mode for integer-to-FP conversions
// must be /SUI or nothing; /U and /SU are not
// allowed. The full set of rounding modes are
// supported though.
0x3c: decode FP_TRAPMODE {
0,7: cvtqs({{ Fc.sf = Fb.sq; }});
}
0x3e: decode FP_TRAPMODE {
0,7: cvtqt({{ Fc = Fb.sq; }});
}
}
}
}
}
// misc FP operate
0x17: decode FP_FULLFUNC {
format BasicOperateWithNopCheck {
0x010: cvtlq({{
Fc.sl = (Fb.uq<63:62> << 30) | Fb.uq<58:29>;
}});
0x030: cvtql({{
Fc.uq = (Fb.uq<31:30> << 62) | (Fb.uq<29:0> << 29);
}});
// We treat the precise & imprecise trapping versions of
// cvtql identically.
0x130, 0x530: cvtqlv({{
// To avoid overflow, all the upper 32 bits must match
// the sign bit of the lower 32. We code this as
// checking the upper 33 bits for all 0s or all 1s.
uint64_t sign_bits = Fb.uq<63:31>;
if (sign_bits != 0 && sign_bits != mask(33))
fault = Integer_Overflow_Fault;
Fc.uq = (Fb.uq<31:30> << 62) | (Fb.uq<29:0> << 29);
}});
0x020: cpys({{ // copy sign
Fc.uq = (Fa.uq<63:> << 63) | Fb.uq<62:0>;
}});
0x021: cpysn({{ // copy sign negated
Fc.uq = (~Fa.uq<63:> << 63) | Fb.uq<62:0>;
}});
0x022: cpyse({{ // copy sign and exponent
Fc.uq = (Fa.uq<63:52> << 52) | Fb.uq<51:0>;
}});
0x02a: fcmoveq({{ Fc = (Fa == 0) ? Fb : Fc; }});
0x02b: fcmovne({{ Fc = (Fa != 0) ? Fb : Fc; }});
0x02c: fcmovlt({{ Fc = (Fa < 0) ? Fb : Fc; }});
0x02d: fcmovge({{ Fc = (Fa >= 0) ? Fb : Fc; }});
0x02e: fcmovle({{ Fc = (Fa <= 0) ? Fb : Fc; }});
0x02f: fcmovgt({{ Fc = (Fa > 0) ? Fb : Fc; }});
0x024: mt_fpcr({{ FPCR = Fa.uq; }});
0x025: mf_fpcr({{ Fa.uq = FPCR; }});
}
}
// miscellaneous mem-format ops
0x18: decode MEMFUNC {
format WarnUnimpl {
0x8000: fetch();
0xa000: fetch_m();
0xe800: ecb();
}
format MiscPrefetch {
0xf800: wh64({{ EA = Rb & ~ULL(63); }},
{{ xc->writeHint(EA, 64, memAccessFlags); }},
IsMemRef, IsDataPrefetch, IsStore, MemWriteOp,
NO_FAULT);
}
format BasicOperate {
0xc000: rpcc({{
#if FULL_SYSTEM
/* Rb is a fake dependency so here is a fun way to get
* the parser to understand that.
*/
Ra = xc->readIpr(AlphaISA::IPR_CC, fault) + (Rb & 0);
#else
Ra = curTick;
#endif
}});
// All of the barrier instructions below do nothing in
// their execute() methods (hence the empty code blocks).
// All of their functionality is hard-coded in the
// pipeline based on the flags IsSerializing,
// IsMemBarrier, and IsWriteBarrier. In the current
// detailed CPU model, the execute() function only gets
// called at fetch, so there's no way to generate pipeline
// behavior at any other stage. Once we go to an
// exec-in-exec CPU model we should be able to get rid of
// these flags and implement this behavior via the
// execute() methods.
// trapb is just a barrier on integer traps, where excb is
// a barrier on integer and FP traps. "EXCB is thus a
// superset of TRAPB." (Alpha ARM, Sec 4.11.4) We treat
// them the same though.
0x0000: trapb({{ }}, IsSerializing, No_OpClass);
0x0400: excb({{ }}, IsSerializing, No_OpClass);
0x4000: mb({{ }}, IsMemBarrier, MemReadOp);
0x4400: wmb({{ }}, IsWriteBarrier, MemWriteOp);
}
#if FULL_SYSTEM
format BasicOperate {
0xe000: rc({{
Ra = xc->readIntrFlag();
xc->setIntrFlag(0);
}}, IsNonSpeculative);
0xf000: rs({{
Ra = xc->readIntrFlag();
xc->setIntrFlag(1);
}}, IsNonSpeculative);
}
#else
format FailUnimpl {
0xe000: rc();
0xf000: rs();
}
#endif
}
#if FULL_SYSTEM
0x00: CallPal::call_pal({{
if (!palValid ||
(palPriv
&& xc->readIpr(AlphaISA::IPR_ICM, fault) != AlphaISA::mode_kernel)) {
// invalid pal function code, or attempt to do privileged
// PAL call in non-kernel mode
fault = Unimplemented_Opcode_Fault;
}
else {
// check to see if simulator wants to do something special
// on this PAL call (including maybe suppress it)
bool dopal = xc->simPalCheck(palFunc);
if (dopal) {
AlphaISA::swap_palshadow(&xc->xcBase()->regs, true);
xc->setIpr(AlphaISA::IPR_EXC_ADDR, NPC);
NPC = xc->readIpr(AlphaISA::IPR_PAL_BASE, fault) + palOffset;
}
}
}}, IsNonSpeculative);
#else
0x00: decode PALFUNC {
format EmulatedCallPal {
0x00: halt ({{
SimExit(curTick, "halt instruction encountered");
}}, IsNonSpeculative);
0x83: callsys({{
xc->syscall();
}}, IsNonSpeculative);
// Read uniq reg into ABI return value register (r0)
0x9e: rduniq({{ R0 = Runiq; }});
// Write uniq reg with value from ABI arg register (r16)
0x9f: wruniq({{ Runiq = R16; }});
}
}
#endif
#if FULL_SYSTEM
format HwLoad {
0x1b: decode HW_LDST_QUAD {
0: hw_ld({{ EA = (Rb + disp) & ~3; }}, {{ Ra = Mem.ul; }}, L);
1: hw_ld({{ EA = (Rb + disp) & ~7; }}, {{ Ra = Mem.uq; }}, Q);
}
}
format HwStore {
0x1f: decode HW_LDST_COND {
0: decode HW_LDST_QUAD {
0: hw_st({{ EA = (Rb + disp) & ~3; }},
{{ Mem.ul = Ra<31:0>; }}, L);
1: hw_st({{ EA = (Rb + disp) & ~7; }},
{{ Mem.uq = Ra.uq; }}, Q);
}
1: FailUnimpl::hw_st_cond();
}
}
format HwMoveIPR {
0x19: hw_mfpr({{
// this instruction is only valid in PAL mode
if (!xc->inPalMode()) {
fault = Unimplemented_Opcode_Fault;
}
else {
Ra = xc->readIpr(ipr_index, fault);
}
}});
0x1d: hw_mtpr({{
// this instruction is only valid in PAL mode
if (!xc->inPalMode()) {
fault = Unimplemented_Opcode_Fault;
}
else {
xc->setIpr(ipr_index, Ra);
if (traceData) { traceData->setData(Ra); }
}
}});
}
format BasicOperate {
0x1e: hw_rei({{ xc->hwrei(); }}, IsSerializing);
// M5 special opcodes use the reserved 0x01 opcode space
0x01: decode M5FUNC {
0x00: arm({{
AlphaPseudo::arm(xc->xcBase());
}}, IsNonSpeculative);
0x01: quiesce({{
AlphaPseudo::quiesce(xc->xcBase());
}}, IsNonSpeculative);
0x10: ivlb({{
AlphaPseudo::ivlb(xc->xcBase());
}}, No_OpClass, IsNonSpeculative);
0x11: ivle({{
AlphaPseudo::ivle(xc->xcBase());
}}, No_OpClass, IsNonSpeculative);
0x20: m5exit_old({{
AlphaPseudo::m5exit_old(xc->xcBase());
}}, No_OpClass, IsNonSpeculative);
0x21: m5exit({{
AlphaPseudo::m5exit(xc->xcBase());
}}, No_OpClass, IsNonSpeculative);
0x30: initparam({{ Ra = xc->xcBase()->cpu->system->init_param; }});
0x40: resetstats({{
AlphaPseudo::resetstats(xc->xcBase());
}}, IsNonSpeculative);
0x41: dumpstats({{
AlphaPseudo::dumpstats(xc->xcBase());
}}, IsNonSpeculative);
0x42: dumpresetstats({{
AlphaPseudo::dumpresetstats(xc->xcBase());
}}, IsNonSpeculative);
0x43: m5checkpoint({{
AlphaPseudo::m5checkpoint(xc->xcBase());
}}, IsNonSpeculative);
0x50: m5readfile({{
AlphaPseudo::readfile(xc->xcBase());
}}, IsNonSpeculative);
0x51: m5break({{
AlphaPseudo::debugbreak(xc->xcBase());
}}, IsNonSpeculative);
0x52: m5switchcpu({{
AlphaPseudo::switchcpu(xc->xcBase());
}}, IsNonSpeculative);
0x53: m5addsymbol({{
AlphaPseudo::addsymbol(xc->xcBase());
}}, IsNonSpeculative);
}
}
#endif
}
Reference in a new issue View git blame Copy permalink