Age | Commit message (Collapse) | Author |
|
The ARM TLBs have a bootUncacheability flag used to make some loads
and stores become uncacheable when booting in FS mode. Later the
flag is cleared to let those loads and stores operate as normal. When
doing a takeOverFrom(), this flag's state is not preserved and is
momentarily reset until the CPSR is touched. On single core runs this
is a non-issue. On multi-core runs this can lead to crashes on the O3
CPU model from the following series of events:
1) takeOverFrom executed to switch from Atomic -> O3
2) All bootUncacheability flags are reset to true
3) Core2 tries to execute a load covered by bootUncacheability, it
is flagged as uncacheable
4) Core2's load needs to replay due to a pipeline flush
3) Core1 core does an action on CPSR
4) The handling code for CPSR then checks all other cores
to determine if bootUncacheability can be set to false
5) Asynchronously set bootUncacheability on all cores to false
6) Core2 replays load previously set as uncacheable and notices
it is now flagged as cacheable, leads to a panic.
This patch implements takeOverFrom() functionality for the ARM TLBs
to preserve flag values when switching from atomic -> detailed.
|
|
in the TLB
Some architectures (currently only x86) require some fixing-up of
physical addresses after a normal address translation. This is usually
to remap devices such as the APIC, but could be used for other memory
mapped devices as well. When running the CPU in a using hardware
virtualization, we still need to do these address fix-ups before
inserting the request into the memory system. This patch moves this
patch allows that code to be used by such CPUs without doing full
address translations.
|
|
This changeset inserts a TLB flush in BaseCPU::switchOut to prevent
stale translations when doing repeated switching. Additionally, the
TLB flushing functionality is exported to the Python to make debugging
of switching/checkpointing easier.
A simulation script will typically use the TLB flushing functionality
to generate a reference trace. The following sequence can be used to
simulate a handover (this depends on how drain is implemented, but is
generally the case) between identically configured CPU models:
m5.drain(test_sys)
[ cpu.flushTLBs() for cpu in test_sys.cpu ]
m5.resume(test_sys)
The generated trace should normally be identical to a trace generated
when switching between identically configured CPU models or
checkpointing and resuming.
|
|
This patch adds support for the memInvalidate() drain method. TLB
flushing is requested by calling the virtual flushAll() method on the
TLB.
Note: This patch renames invalidateAll() to flushAll() on x86 and
SPARC to make the interface consistent across all supported
architectures.
|
|
This patch addresses the comments and feedback on the preceding patch
that reworks the clocks and now more clearly shows where cycles
(relative cycle counts) are used to express time.
Instead of bumping the existing patch I chose to make this a separate
patch, merely to try and focus the discussion around a smaller set of
changes. The two patches will be pushed together though.
This changes done as part of this patch are mostly following directly
from the introduction of the wrapper class, and change enough code to
make things compile and run again. There are definitely more places
where int/uint/Tick is still used to represent cycles, and it will
take some time to chase them all down. Similarly, a lot of parameters
should be changed from Param.Tick and Param.Unsigned to
Param.Cycles.
In addition, the use of curTick is questionable as there should not be
an absolute cycle. Potential solutions can be built on top of this
patch. There is a similar situation in the o3 CPU where
lastRunningCycle is currently counting in Cycles, and is still an
absolute time. More discussion to be had in other words.
An additional change that would be appropriate in the future is to
perform a similar wrapping of Tick and probably also introduce a
Ticks class along with suitable operators for all these classes.
|
|
Making the CheckerCPU a runtime time option requires the code to be compatible
with ISAs other than ARM. This patch adds the appropriate function
stubs to allow compilation.
|
|
|
|
|
|
|
|
--HG--
rename : src/sim/fault.hh => src/sim/fault_fwd.hh
|
|
|
|
Also move the "Fault" reference counted pointer type into a separate file,
sim/fault.hh. It would be better to name this less similarly to sim/faults.hh
to reduce confusion, but fault.hh matches the name of the type. We could change
Fault to FaultPtr to match other pointer types, and then changing the name of
the file would make more sense.
|
|
|
|
|
|
the timing simple CPU to use it.
|
|
|
|
The translate function simplifies code and removes some compiler
warnings in gcc 3.4
|
|
--HG--
extra : convert_revision : cc0e62a5a337fd5bf332ad33bed61c0d505a936f
|
|
SimObjects not yet updated:
- Process and subclasses
- BaseCPU and subclasses
The SimObject(const std::string &name) constructor was removed. Subclasses
that still rely on that behavior must call the parent initializer as
: SimObject(makeParams(name))
--HG--
extra : convert_revision : d6faddde76e7c3361ebdbd0a7b372a40941c12ed
|
|
--HG--
extra : convert_revision : a04a30df0b6246e877a1cea35420dbac94b506b1
|
|
Each "strand" may need to have a private copy of this state, but I couldn't
find anywhere in the spec that said that after looking briefly.
This prevents writes to the thread context in o3 which was causing the
pipeline to be flushed and stopping any forward progress. The other ASI
accessible state will probably need to be accessed differently if/when we get
O3 full system up and running.
--HG--
extra : convert_revision : fa7fba812d7f76564ef4a23818e60f536710d557
|
|
src/arch/alpha/vtophys.cc:
src/arch/alpha/vtophys.hh:
src/arch/sparc/arguments.hh:
move Copy* to vport since it's generic for all the ISAs
src/arch/sparc/isa_traits.hh:
the Solaris kernel sets up a virtual-> real mapping for all memory starting at SegKPMBase
src/arch/sparc/pagetable.hh:
add a class for getting bits out of the TteTag
src/arch/sparc/remote_gdb.cc:
add 32bit support kinda.... If its 32 bit
src/arch/sparc/remote_gdb.hh:
Add 32bit register offsets too.
src/arch/sparc/tlb.cc:
cleanup generation of tsb pointers
src/arch/sparc/tlb.hh:
add function to return tsb pointers for an address
make lookup public so vtophys can use it
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
write vtophys for sparc
src/base/bitfield.hh:
return a mask of bits first->last
src/mem/vport.cc:
src/mem/vport.hh:
move Copy* here since it's ISA generic
--HG--
extra : convert_revision : c42c331e396c0d51a2789029d8e232fe66995d0f
|
|
Make our replacement algorithm same as legion (although not same as the spec)
itb should be 64 entries not 48
src/arch/sparc/tlb.cc:
Bug fixes in the TLB
Make our replacement algorithm same as legion (although not same as the spec)
src/arch/sparc/tlb.hh:
Make our replacement algorithm same as legion (although not same as the spec)
src/python/m5/objects/SparcTLB.py:
itb should be 64 entries too
--HG--
extra : convert_revision : 1b5cb3597091e3cfe293e94f6f2219b1e621c35f
|
|
into zower.eecs.umich.edu:/eecshome/m5/sparcfs
--HG--
extra : convert_revision : 2764b356ef01d1fcb6ed272e4ef96179cd651d4e
|
|
--HG--
extra : convert_revision : 39e2638a10bf3e821e8f3d4d8c664008c98fc921
|
|
--HG--
extra : convert_revision : f79f863393f918ff9363b2c261f8c0dfec64312e
|
|
Fix fault formating and code for traps
fix a couple of bugs in the decoder
Cleanup/fix page table entry code
Implement more mmaped iprs, fix numbered tlb insertion code, add function to dump tlb contents
Don't panic if we differ from legion on a tcc instruction because of where legion prints its data and where we print our data
src/arch/sparc/faults.cc:
Fix fault formating and code for traps
src/arch/sparc/intregfile.hh:
allocate the correct number of global registers
src/arch/sparc/isa/decoder.isa:
fix a couple of bugs in the decoder: wrasi should write asi not ccr, done/retry should get hpstate from htstate
src/arch/sparc/pagetable.hh:
cleanup/fix page table code
src/arch/sparc/tlb.cc:
implement more mmaped iprs, fix numbered insertion code, add function to dump tlb contents
src/arch/sparc/tlb.hh:
add functions to write TagAccess register on tlb miss and to dump all tlb entries for debugging
src/cpu/exetrace.cc:
dump tlb entries on error, don't consider differences the cycle we take a trap to be bad.
--HG--
extra : convert_revision : d7d771900f6f25219f3dc6a6e51986d342a32e03
|
|
src/arch/SConscript:
add mmaped_ipr.hh to switch headers
src/arch/sparc/asi.hh:
make ASI_IMPLICT=0 so by default nothing needs to be done
src/arch/sparc/miscregfile.hh:
miscregfile no longer needs to include asi.hh
src/arch/sparc/tlb.cc:
src/arch/sparc/tlb.hh:
implement panic instructions for mmaped ipr reads
src/cpu/simple/atomic.cc:
add check for mmaped iprs and handle them if it exists
src/mem/request.hh:
allocate space in the flags for mmaped iprs. Put in in the first 8 bits so that by default its fast. Move the other flags up 8 bits
--HG--
extra : convert_revision : 31255b0494588c4d06a727fe35241121d741b115
|
|
src/arch/sparc/SConscript:
Add code to serialize/unserialze tlb entries
src/arch/sparc/asi.cc:
src/arch/sparc/asi.hh:
update asi names for how they're listed in the supplement
add asis
add more asi functions
src/arch/sparc/isa_traits.hh:
move the interrupt stuff and some basic address space stuff into isa traits
src/arch/sparc/miscregfile.cc:
src/arch/sparc/miscregfile.hh:
add mmu registers to tlb
get rid of implicit asi stuff... the tlb will handle it
src/arch/sparc/regfile.hh:
make isnt/dataAsid return ints not asis
src/arch/sparc/tlb.cc:
src/arch/sparc/tlb.hh:
first cut at sparc tlb
src/arch/sparc/vtophys.hh:
pagatable nedes to be included here
src/mem/request.hh:
add asi and if the request is a memory mapped register to the requset object
src/sim/host.hh:
fix incorrect definition of LL
--HG--
extra : convert_revision : 6c85cd1681c62c8cd8eab04f70b1f15a034b0aa3
|
|
use memcpy instead of bcopy
s/u_int32_t/uint32_t/g
fixup endian code to work with solaris
hack to make sure htole() works... Nate, have a good idea to fix this?
src/arch/sparc/faults.cc:
set the reset address to be 40 bits. Makes PC printing easier at least for now.
src/arch/sparc/isa/base.isa:
fix endian issues with condition codes
src/arch/sparc/tlb.hh:
add implemented physical addres constants
src/arch/sparc/utility.hh:
add tlb.hh to utilities
src/base/loader/raw_object.cc:
add a symbol <filename>_start to the symbol table for binaries files
src/base/remote_gdb.cc:
use memcpy instead of bcopy
src/cpu/exetrace.cc:
clean up printing a bit more
src/cpu/m5legion_interface.h:
add tons to the shared interface
src/dev/ethertap.cc:
s/u_int32_t/uint32_t/g
src/dev/ide_atareg.h:
fixup endian code to work with solaris
src/dev/pcidev.cc:
src/sim/param.hh:
hack to make sure htole() works...
--HG--
extra : convert_revision : 4579392184b40bcc1062671a953c6595c685e9b2
|
|
the virtual address.
--HG--
extra : convert_revision : 41478abc4d21d504420f6842338675c0767f7cf9
|
|
--HG--
extra : convert_revision : ba08da78693cc6f59f7358134f121f471910dbf6
|
|
file functions to not take faults
--HG--
extra : convert_revision : 1cef0734462ee2e4db12482462c2ab3c134d3675
|
|
setup all initialization stuff for UA2005
Setup fullsys build options
Start to make fullsystem compile
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
RCS to BitKeeper
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Add support for doing virtual to physical translation using the in-memory
page table
src/arch/sparc/vtophys.cc:
a tad bit of error checking
src/arch/sparc/vtophys.hh:
Cleanup of full-system global variables, primarily in simple_cpu.cc,
to allow multi-system simulations. (Multiple systems not yet yested
though.) Also changes to build sim_smt in full-system mode (though
with only SimpleCPU and not the full timing CPU for now).
Still to do: minimize changes in SimpleCPU code between full-system
and application modes... way too many ifdefs there. Much of the
full-system stuff moved into SimpleCPU should be put in a new System
object to allow multiprocessor simulations.
Converted last remaining modules from C to C++ (mostly in /old).
Renamed all .c files to .cc and a few .h files to .hh.
Renamed architecture-specific files in arch/$TARGET from
$TARGET.{cc,hh,def} to machine.{cc,hh,def} to get rid of pointless
intermediate files in object directory. Split exo-specific
definitions out of machine.hh into machine_exo.h.
Specifics:
In machine.def, null resource descriptors must be FUClamd_NA (and not
NA) to pass C++ type checking.
Enhanced error checking/reporting in bas
src/arch/sparc/vtophys.cc:
- Get rid of my String class, the Vector class, the bitvector class, and my
doubly linked list class.
- Convert tokenize, to_number (formerly StringToNumber) and eat_white to
function on stl strings.
- Change most cases of char * and const char * to string, or const string &
- Some formatting and style nits, but not too many.
src/arch/sparc/vtophys.cc:
simplify
src/arch/sparc/vtophys.cc:
Renamed SimpleCPU::(read|write)_(byte|half|word|qword) to just read &
write, overloaded on the type of the 'data' argument. Merged the
full-system and non-full-system implementations of these eight
original functions into two common template functions.
To support this, also renamed (read|write)[1248] on memory_object and
derivatives to just read & write, again overloaded on the type of the
'data' argument. Many of these functions could now be condensed into
a few template functions (though with a level of indirection so that
the interface can remain virtual). I did not do that though.
src/arch/sparc/vtophys.cc:
First pass at compiling with gcc 3.x. Lots of "std::" in header files,
"using namespace std" in source files. (Note policy of not putting "using"
statements in headers or before includes in sources.)
Still not able to compile with gcc 3.2.
Errors:
- Can't create an ifstream from a file descriptor anymore (breaks IniFile).
- "`class MSHR::MSHRegister' is private" errors in mshr.cc and prefetch_cache.cc:
not clear why since it's in the public part of the class declaration.
- cpu.cc:879: can't match a reference and 0 (specifically "no match for `bool ?
SimObjectParam<PipeTrace*>& : int' operator")
- pipetrace.cc: "invalid conversion from `int' to `std::_Ios_Fmtflags'"
Warnings:
- strstream now deprecated... needs some rewriting in sat_counter.hh
and hybrid_pred.hh (need to get all that code out of the headers anyway)
- trace.hh macro problem: cpp now says 'pasting "::" and "Event" does not
give a valid preprocessing token'
- major "implicit typename" issues in base/sized.hh
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Implement some interval statistics for full system mode.
Create a callpal function that is called when a callpal occurs so it's easier
to manipulate the statics.
Rework the vtophys stuff to make it a bit cleaner.
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Get rid of almost all old-style object names. This commit is equivalent
to running the following script on the current head:
#! /bin/sh
find \( -name '*.cc' -o -name '*.hh' \) -exec perl -pi -e '\
s/\bmemory_object\b(?!\.hh)/FunctionalMemory/g;\
s/\bvirtual_memory\b(?!\.hh)/VirtualMemory/g;\
s/\bmain_memory\b(?!\.hh)/MainMemory/g;\
s/\bphysical_memory\b(?!\.hh)/PhysicalMemory/g;\
s/\bspec_memory\b(?!\.hh)/SpeculativeMemory/g;\
s/\bMemObj\b(?!\.hh)/TimingMemObj/g;\
s/\bmemory_translation\b(?!\.hh)/AddressTranslator/g;\
s/\balpha_tlb\b(?!\.hh)/AlphaTlb/g;\
s/\balpha_itb\b(?!\.hh)/AlphaItb/g;\
s/\balpha_dtb\b(?!\.hh)/AlphaDtb/g;\
s/\bmemory_controller\b(?!\.hh)/MemoryController/g;\
s/\bstorebuffer_t\b(?!\.hh)/StoreBuffer/g;\
s/\bstorebuffer_entry_t\b(?!\.hh)/StoreBufferEntry/g;\
s/\bcreate_vector_t\b(?!\.hh)/CreateVector/g;\
s/\bcv_spec_state\b(?!\.hh)/CreateVecSpecState/g;\
s/\bspec_state_list\b(?!\.hh)/SpecStateList/g;\
s/\bdyn_inst_t\b(?!\.hh)/DynInst/g;' {} \;
src/arch/sparc/vtophys.cc:
since cprintf properly deals with 64-bit types, stop using FMT* as much as
possible
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Change byte_t etc. to C99 standard int8_t etc.
Other than old/host.h, all other changes were produced by this script:
#! /bin/sh
find \( -name '*.cc' -o -name '*.hh' -o -name '*.c' -o -name '*.h' -o -name 'machine.def' \) -exec perl -pi -e '\
s/\bbyte_t\b(?!\.hh)/uint8_t/g;\
s/\bsbyte_t\b(?!\.hh)/int8_t/g;\
s/\bhalf_t\b(?!\.hh)/uint16_t/g;\
s/\bshalf_t\b(?!\.hh)/int16_t/g;\
s/\bword_t\b(?!\.hh)/uint32_t/g;\
s/\bsword_t\b(?!\.hh)/int32_t/g;\
s/\bqword_t\b(?!\.hh)/uint64_t/g;\
s/\bsqword_t\b(?!\.hh)/int64_t/g;\
s/\bbool_t\b(?!\.hh)/bool/g;\
s/\bdfloat_t\b(?!\.hh)/double/g;\
s/\bsfloat_t\b(?!\.hh)/float/g;' {} \;
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Add CVS Id tags
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Massive renaming to (almost) eliminate all md_* and MD_* names
in preparation for total exorcism of machine.def.
Most of the changes in this commit were performed with the
following perl script (perl -pi <script> <files>). A small
amount of manual fixup was needed to (mostly getting rid of
the Addr typedefs in the various memory objects now that
the former md_addr_t has that name).
# rename machine-dependent types and constants (will be moving into ISA traits object)
s/md_addr_t/Addr/g;
s/md_intreg_t/IntReg/g;
s/md_gpr_t/IntRegFile/g;
s/md_fpreg_t/FloatReg/g;
s/md_fpr_t/FloatRegFile/g;
s/md_ctrlreg_t/MiscReg/g;
s/md_ctrl_t/MiscRegFile/g;
s/md_ipr_t/InternalProcReg/g;
s/md_anyreg_t/AnyReg/g;
s/md_inst_t/MachInst/g;
s/regs_t/RegFile/g;
# manually fix declaration in old/regs.h and a few forward decls
s/struct RegFile/RegFile/g;
s/MD_NUM_IREGS/NumIntRegs/g;
s/MD_NUM_FREGS/NumFloatRegs/g;
s/MD_NUM_CREGS/NumMiscRegs/g;
s/MD_IPR_NUM/NumInternalProcRegs/g;
s/MD_TOTAL_REGS/TotalNumRegs/g;
s/MD_REG_ZERO/ZeroReg/g;
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
New ISA description system. No more machine.def!
Instructions are now decoded into StaticInst objects, and all static
instruction properties (including execution behavior) are associated
with those objects. Extended documentation in progress.
Currently supports Alpha only; PISA will not compile.
Use END_OF_MACHINE_DOT_DEF tag to extract previous version.
src/arch/sparc/vtophys.cc:
get rid of MD_IPR_foo and call it IPR_foo
add some comments to describe what the various PALtemp registers do
formatting
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
license
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
a little style
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Add attribution to license.
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Make include paths explicit.
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
CopyData and CopyString moved from kernel.cc to vtophys.cc
kernel.cc and kernel.hh moved to kern/tru64
src/arch/sparc/vtophys.hh:
Include isa_traits.hh for Addr
src/arch/sparc/vtophys.cc:
formatting fixes
src/arch/sparc/vtophys.cc:
fix up vtophys to deal with translations if there
is no ptbr, and to deal with PAL addresses
add ptomem which is just a wrapper for dma_addr
src/arch/sparc/vtophys.hh:
add ptomem which is a wrapper for dma_addr with the
same usage as vtomem
src/arch/sparc/vtophys.cc:
Fix to remote debugger while in PAL code
src/arch/sparc/vtophys.cc:
Remote an old hack that is now unnecessary
src/arch/sparc/vtophys.cc:
Removed buggy code that tries to fix PAL addresses (may cause problems
while trying to debug in PAL code, but that should do this fix outside
of vtophys)
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Updated Copyright
src/arch/sparc/vtophys.cc:
added back some code andrew removed and couldn't remember why.
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
rename CopyData to CopyOut and implement CopyIn to copy data
from the simulator into the simulatee
src/arch/sparc/vtophys.cc:
fixed a bad merge from linux<->tru64
src/arch/sparc/vtophys.cc:
Check max address pal can be at so we don't do the wrong conversion
if gdb asks for an unaligned access.
src/arch/sparc/vtophys.cc:
PGOFSET -> ALPHA_PGOFSET to avoid include file problems
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
in the arch/alpha directory we should use arch/alpha, not
targetarch. sort includes while we're here.
src/arch/sparc/vtophys.cc:
use new constants, functions and structs to clean up the
vtophys code.
src/arch/sparc/vtophys.hh:
Clean up a little bit and make the protypes match new changes.
src/arch/sparc/vtophys.cc:
deal with isa addition
src/arch/sparc/vtophys.cc:
shuffle files around for new directory structure
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Remove RCS Id string
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
Update copyright dates and author list
src/arch/sparc/vtophys.cc:
Added a using directive for AlphaISA
src/arch/sparc/vtophys.hh:
Added the AlphaISA namespace specifier where needed
src/arch/sparc/vtophys.hh:
Made Addr a global type
src/arch/sparc/vtophys.cc:
Change access to the IPR to go through the XC.
src/arch/sparc/vtophys.cc:
Avoid directly accessing objects within the XC.
src/arch/sparc/vtophys.cc:
src/arch/sparc/vtophys.hh:
fixed for new memory system
put in namespace AlphaISA
src/arch/alpha/vtophys.cc:
src/arch/alpha/vtophys.hh:
Remove authors from copyright.
src/arch/sparc/vtophys.cc:
bk cp alpha/vtophys.cc sparc/vtophys.cc
src/arch/sparc/vtophys.hh:
bk cp alpha/vtophys.hh sparc/vtophys.hh
src/arch/sparc/SConscript:
remove fullsystem files that don't exist
src/arch/sparc/isa_traits.hh:
split off fullsystem and se iprs into two functions to remove lots of #ifs
src/arch/sparc/regfile.hh:
split off fullsystem and se iprs into two functions to remove lots of #ifs
setup all initialization stuff for UA2005
src/arch/sparc/system.cc:
src/arch/sparc/system.hh:
Add system level tick storage to make stick be syncronized across multiple processors
src/arch/sparc/vtophys.hh:
start to create a vtophys for Sparc
src/base/loader/symtab.hh:
Addr is defined in sim/host.hh
--HG--
extra : convert_revision : ae4658e6bc2e2b51d7fb1c711d069ffe99e1377d
|