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.
Even though we're not incorrect about operator precedence, let's add
some parens in some particularly confusing places to placate GCC 4.3
so that we don't have to turn the warning off. Agreed that this is a
bit of a pain for those users who get the order of operations correct,
but it is likely to prevent bugs in certain cases.
In O3, a nop is used to carry faults down the pipeline that didn't originate
from an instruction. If the instruction doesn't do anything, that is just
returns NoFault, but doesn't have IsNop set, the NoFault will overwrite the
fault that's being sent down and nothing will happen.
--HG--
extra : convert_revision : 54d99002b550ca0e1cf14603f588dc1038e3e535
into doughnut.hpl.hp.com:/home/gblack/newmem-o3-micro
src/cpu/base_dyn_inst_impl.hh:
src/cpu/o3/fetch_impl.hh:
Hand merge
--HG--
extra : convert_revision : 0c0692033ac30133672d8dfe1f1a27e9d9e95a3d
src/arch/x86/isa/macroop.isa:
Make microOp vs microop and macroOp vs macroop capitilization consistent. Also fill out the emulation environment handling a little more, and use an object to pass around output code.
src/arch/x86/isa/microops/base.isa:
Make microOp vs microop and macroOp vs macroop capitilization consistent. Also adjust python to C++ bool translation.
--HG--
extra : convert_revision : 6f4bacfa334c42732c845f9a7f211cbefc73f96f
(which defines fenv) doesn't necessarily extend to c++ and it is a problem with solaris. If really
desired this could wrap the ieeefp interface found in bsd* as well, but I see no need at the moment.
src/arch/alpha/isa/fp.isa:
src/arch/sparc/isa/formats/basic.isa:
use m5_fesetround()/m5_fegetround() istead of fenv interface directly
src/arch/sparc/isa/includes.isa:
use base/fenv instead of fenv directly
src/base/SConscript:
add fenv to sconscript
src/base/fenv.hh:
src/base/random.cc:
m5 implementation to standerdize fenv across platforms.
--HG--
extra : convert_revision : 38d2629affd964dcd1a5ab0db4ac3cb21438e72c
Add support for a twin 64 bit int load
Add Memory barrier and write barrier flags as appropriate
Make atomic memory ops atomic
src/arch/alpha/isa/mem.isa:
src/arch/alpha/locked_mem.hh:
src/cpu/base_dyn_inst.hh:
src/mem/cache/cache_blk.hh:
src/mem/cache/cache_impl.hh:
rename store conditional stuff as extra data so it can be used for conditional swaps as well
src/arch/alpha/types.hh:
src/arch/mips/types.hh:
src/arch/sparc/types.hh:
add a largest read data type for statically allocating read buffers in atomic simple cpu
src/arch/isa_parser.py:
Add support for a twin 64 bit int load
src/arch/sparc/isa/decoder.isa:
Make atomic memory ops atomic
Add Memory barrier and write barrier flags as appropriate
src/arch/sparc/isa/formats/mem/basicmem.isa:
add post access code block and define a twinload format for twin loads
src/arch/sparc/isa/formats/mem/blockmem.isa:
remove old microcoded twin load coad
src/arch/sparc/isa/formats/mem/mem.isa:
swap.isa replaces the code in loadstore.isa
src/arch/sparc/isa/formats/mem/util.isa:
add a post access code block
src/arch/sparc/isa/includes.isa:
need bigint.hh for Twin64_t
src/arch/sparc/isa/operands.isa:
add a twin 64 int type
src/cpu/simple/atomic.cc:
src/cpu/simple/atomic.hh:
src/cpu/simple/base.hh:
src/cpu/simple/timing.cc:
add support for twinloads
add support for swap and conditional swap instructions
rename store conditional stuff as extra data so it can be used for conditional swaps as well
src/mem/packet.cc:
src/mem/packet.hh:
Add support for atomic swap memory commands
src/mem/packet_access.hh:
Add endian conversion function for Twin64_t type
src/mem/physical.cc:
src/mem/physical.hh:
src/mem/request.hh:
Add support for atomic swap memory commands
Rename sc code to extradata
--HG--
extra : convert_revision : 69d908512fb34a4e28b29a6e58b807fb1a6b1656
fix unaligned accesses in mmaped disk device
src/arch/sparc/isa/decoder.isa:
get (ld|st)fsr ops working right. In reality the fp enable check needs to go higher up in the emitted code
src/arch/sparc/isa/formats/basic.isa:
move the cexec into the aexec field
src/cpu/exetrace.cc:
copy the exception state from legion when we get it wrong. We aren't going to get it right without an fp emulation layer
src/dev/sparc/mm_disk.cc:
src/dev/sparc/mm_disk.hh:
fix unaligned accesses in the memory mapped disk device
--HG--
extra : convert_revision : aaa33096b08cf0563fe291d984a87493a117e528
src/arch/sparc/floatregfile.cc:
fix fp read/writing to registers... looking for suggestions on cleaner ways if anyone has them
src/arch/sparc/isa/decoder.isa:
fix some fp implementations
src/arch/sparc/isa/formats/basic.isa:
add new fp op class that 0 cexec in fsr and sets rounding mode for the up comming op
src/arch/sparc/isa/includes.isa:
include the appropriate header files for the rounding code
src/arch/sparc/miscregfile.cc:
print fsr out when it's read/written and the Sparc traceflgas in on
src/cpu/exetrace.cc:
fix printing of float registers
--HG--
extra : convert_revision : 49faab27f2e786a8455f9ca0f3f0132380c9d992
src/arch/sparc/isa/base.isa:
Added passesFpCondition function to help with fbfcc and fbpfcc instructions.
src/arch/sparc/isa/decoder.isa:
Added fbfcc and fbpfcc instructions, and cleaned up branch code slightly.
src/arch/sparc/isa/formats/branch.isa:
Minor cleanup.
--HG--
extra : convert_revision : 6586b46418f1f70bace41407f267fee30c657714
into zower.eecs.umich.edu:/eecshome/m5/newmem
src/arch/sparc/isa/formats/mem/util.isa:
src/arch/sparc/isa_traits.hh:
src/arch/sparc/system.cc:
Hand Merge
--HG--
extra : convert_revision : d5e0c97caebb616493e2f642e915969d7028109c
some fixes to fp instructions to use the single precision registers
if this is an fp op emit fp check code
add fpregs to m5legion struct
src/arch/sparc/floatregfile.cc:
Make Sparc traceflag even more chatty
src/arch/sparc/isa/base.isa:
add code to check if the fpu is enabled
src/arch/sparc/isa/decoder.isa:
some fixes to fp instructions to use the single precision registers
fix smul again
fix subc/subcc/subccc condition code setting
src/arch/sparc/isa/formats/basic.isa:
src/arch/sparc/isa/formats/mem/util.isa:
if this is an fp op emit fp check code
src/cpu/exetrace.cc:
check fp regs as well as int regs
src/cpu/m5legion_interface.h:
add fpregs to m5legion struct
--HG--
extra : convert_revision : e7d26d10fb8ce88f96e3a51f84b48c3b3ad2f232
pretty close to compiling w/ suns compiler
briefly:
add dummy return after panic()/fatal()
split out flags by compiler vendor
include cstring and cmath where appropriate
use std namespace for string ops
SConstruct:
Add code to detect compiler and choose cflags based on detected compiler
Fix zlib check to work with suncc
src/SConscript:
split out flags by compiler vendor
src/arch/sparc/isa/decoder.isa:
use correct namespace for sqrt
src/arch/sparc/isa/formats/basic.isa:
add dummy return around panic
src/arch/sparc/isa/formats/integerop.isa:
use correct namespace for stringops
src/arch/sparc/isa/includes.isa:
include cstring and cmath where appropriate
src/arch/sparc/isa_traits.hh:
remove dangling comma
src/arch/sparc/system.cc:
dummy return to make sun cc front end happy
src/arch/sparc/tlb.cc:
src/base/compression/lzss_compression.cc:
use std namespace for string ops
src/arch/sparc/utility.hh:
no reason to say something is unsigned unsigned int
src/base/compression/null_compression.hh:
dummy returns to for suncc front end
src/base/cprintf.hh:
use standard variadic argument syntax instead of gnuc specefic renaming
src/base/hashmap.hh:
don't need to define hash for suncc
src/base/hostinfo.cc:
need stdio.h for sprintf
src/base/loader/object_file.cc:
munmap is in std namespace not null
src/base/misc.hh:
use M5 generic noreturn macros
use standard variadic macro __VA_ARGS__
src/base/pollevent.cc:
we need file.h for file flags
src/base/random.cc:
mess with include files to make suncc happy
src/base/remote_gdb.cc:
malloc memory for function instead of having a non-constant in an array size
src/base/statistics.hh:
use std namespace for floor
src/base/stats/text.cc:
include math.h for rint (cmath won't work)
src/base/time.cc:
use suncc version of ctime_r
src/base/time.hh:
change macro to work with both gcc and suncc
src/base/timebuf.hh:
include cstring from memset and use std::
src/base/trace.hh:
change variadic macros to be normal format
src/cpu/SConscript:
add dummy returns where appropriate
src/cpu/activity.cc:
include cstring for memset
src/cpu/exetrace.hh:
include cstring fro memcpy
src/cpu/simple/base.hh:
add dummy return for panic
src/dev/baddev.cc:
src/dev/pciconfigall.cc:
src/dev/platform.cc:
src/dev/sparc/t1000.cc:
add dummy return where appropriate
src/dev/ide_atareg.h:
make define work for both gnuc and suncc
src/dev/io_device.hh:
add dummy returns where approirate
src/dev/pcidev.hh:
src/mem/cache/cache_impl.hh:
src/mem/cache/miss/blocking_buffer.cc:
src/mem/cache/tags/lru.hh:
src/mem/cache/tags/split.hh:
src/mem/cache/tags/split_lifo.hh:
src/mem/cache/tags/split_lru.hh:
src/mem/dram.cc:
src/mem/packet.cc:
src/mem/port.cc:
include cstring for string ops
src/dev/sparc/mm_disk.cc:
add dummy return where appropriate
include cstring for string ops
src/mem/cache/miss/blocking_buffer.hh:
src/mem/port.hh:
Add dummy return where appropriate
src/mem/cache/tags/iic.cc:
cast hastSets to double for log() call
src/mem/physical.cc:
cast pmemAddr to char* for munmap
src/sim/byteswap.hh:
make define work for suncc and gnuc
--HG--
extra : convert_revision : ef8a1f1064e43b6c39838a85c01aee4f795497bd
Only allow writing/reading of 32 bits of Y
Only allow writing/reading 32 bits of pc when pstate.am
Put any loaded data on the first half of a micro-op in uReg0 so it can't
overwrite the register we are using for address calculation
only erase a entry from the lookup table if it's valid
Put in a temporary check to make sure that lookup table and tlb array stay in sync
if we are interrupted in the middle of a mico-op, reset the micropc/nexpc
so we start on the first part of it when we come back
src/arch/sparc/isa/decoder.isa:
fix smul and sdiv to sign extend, and handle overflow/underflow corretly
Only allow writing/reading of 32 bits of Y
Only allow writing/reading 32 bits of pc when pstate.am
Put any loaded data on the first half of a micro-op in uReg0 so it can't
overwrite the register we are using for address calculation
src/arch/sparc/isa/formats/mem/blockmem.isa:
Put any loaded data on the first half of a micro-op in uReg0 so it can't
overwrite the register we are using for address calculation
src/arch/sparc/isa/includes.isa:
Use limits for 32bit underflow/overflow detection
src/arch/sparc/tlb.cc:
only erase a entry from the lookup table if it's valid
Put in a temporary check to make sure that lookup table and tlb array stay in sync
src/arch/sparc/tlb_map.hh:
add a print function to dump the tlb lookup table
src/cpu/simple/base.cc:
if we are interrupted in the middle of a mico-op, reset the micropc/nexpc
so we start on the first part of it when we come back
--HG--
extra : convert_revision : 50a23837fd888393a5c2aa35cbd1abeebb7f55d4
Increment instruction count on first micro-op instead of last
src/arch/sparc/isa/decoder.isa:
Implement a twin load for ASI_LDTX_P(0xe2)
src/arch/sparc/isa/formats/mem/blockmem.isa:
set the new flag IsFirstMicroOp when needed
src/cpu/simple/atomic.cc:
Increment instruction count on first micro-op instead of last (because if we take a fault on a micro coded instruction it should be counted twice acording to legion)
src/cpu/static_inst.hh:
Add IsFirstMicroop flag to static insts
--HG--
extra : convert_revision : 02bea93d38c03bbafe4570665eb4c01c11caa2fc
bugfixes and demap implementation in tlb
ignore some more differencs for one cycle
src/arch/sparc/isa/formats/mem/blockmem.isa:
twinx has 2 micro-ops
src/arch/sparc/isa/formats/mem/util.isa:
fix the fault check for twinx
src/arch/sparc/tlb.cc:
tlb bugfixes and write demapping code
src/cpu/exetrace.cc:
don't halt on a couple more instruction (ldx, stx) when things differ
beacuse of the way tlb faults are handled in legion.
--HG--
extra : convert_revision : 1e156dead6ebd58b257213625ed63c3793ef4b71
into zower.eecs.umich.edu:/eecshome/m5/newmem
src/arch/sparc/isa/formats/mem/blockmem.isa:
src/arch/sparc/isa/operands.isa:
Hand Merge
--HG--
extra : convert_revision : 4c54544e5c7a61f055ea9b00ccf5f8510df0e6c2
src/arch/sparc/isa/decoder.isa:
Added the extended twin load instructions
src/arch/sparc/isa/formats/mem/blockmem.isa:
Added stuff to implement the extended twin loads. This created alot of duplication which I'll deal with later.
--HG--
extra : convert_revision : 5d8bdaacbfe83d21d3a396ce30ace90aeefc54d8
src/arch/sparc/isa/decoder.isa:
Changed the names of the twin loads to match the 2005 spec. They still use the old format though.
src/arch/sparc/isa/formats/mem/blockmem.isa:
Added code to generate twin loads
src/arch/sparc/isa/formats/mem/util.isa:
Added an alignment check for twin loads
src/arch/sparc/isa/operands.isa:
Comment explaining twin load operands.
--HG--
extra : convert_revision : ad42821a97dcda17744875b1e5dc00a9642e59b7
src/arch/isa_parser.py:
Rearranged things so that classes with more than one execute function treat operands properly.
1. Eliminated the CodeBlock class
2. Created a SubOperandList
3. Redefined how InstObjParams is constructed
To define an InstObjParam, you can either pass in a single code literal which will be named "code", or you can pass in a dictionary of code snippets which will be substituted into the Templates. In order to get this to work, there is a new restriction that each template has only one function in it. These changes should only affect memory instructions which have regular and split execute functions.
Also changed the MiscRegs so that they use the instrunctions srcReg and destReg arrays.
src/arch/sparc/isa/formats/basic.isa:
src/arch/sparc/isa/formats/branch.isa:
src/arch/sparc/isa/formats/integerop.isa:
src/arch/sparc/isa/formats/mem/basicmem.isa:
src/arch/sparc/isa/formats/mem/blockmem.isa:
src/arch/sparc/isa/formats/mem/util.isa:
src/arch/sparc/isa/formats/nop.isa:
src/arch/sparc/isa/formats/priv.isa:
src/arch/sparc/isa/formats/trap.isa:
Rearranged to work with new InstObjParam scheme.
src/cpu/o3/sparc/dyn_inst.hh:
Added functions to access the miscregs using the indexes from instructions srcReg and destReg arrays. Also changed the names of the other accessors so that they have the suffix "Operand" if they use those arrays.
src/cpu/simple/base.hh:
Added functions to access the miscregs using the indexes from instructions srcReg and destReg arrays.
--HG--
extra : convert_revision : c91e1073138b72bcf4113a721e0ed40ec600cf2e