Commit graph

23 commits

Author SHA1 Message Date
Ron Dreslinski
9e3d79694c Move to a model with a unified request object.
Constructor takes a bool to signify that it is either a cpu_request or not a cpu_request.
When accedding variables of a cpu_request it asserts that it is a cpu_request.
It also asserts that a value being read has been written at some point in time prior (not gaurnteeing it is up to date, but it was at least written before read).

There is also a isCpuReq() function to determine if this is a cpu_request.  It should be called before accesing a cpu_request only variable.

SConscript:
    Add compilation support for request.cc
arch/alpha/tlb.cc:
arch/alpha/tlb.hh:
cpu/cpu_exec_context.hh:
cpu/exec_context.hh:
cpu/simple/cpu.cc:
cpu/simple/cpu.hh:
dev/io_device.cc:
mem/page_table.cc:
mem/page_table.hh:
mem/port.cc:
    Update for unified request object and accessor functions.
mem/request.hh:
    Remove CpuRequest, make it a unified object.  Make variables private with accessor functions.

    May want to move things from .cc file into header (usually a assert() and either returning a value, or writting two).

--HG--
extra : convert_revision : f1e45cc490dadc7a418634539b03c3e72684a6e3
2006-04-07 15:54:48 -04:00
Gabe Black
832311a170 Fixed for full system.
--HG--
extra : convert_revision : 28c9cd55d887c9de7156c8cf76b7b91117f749d5
2006-04-06 15:21:52 -04:00
Gabe Black
6a962f8343 Merge m5.eecs.umich.edu:/bk/newmem
into  ewok.(none):/home/gblack/m5/newmem

--HG--
extra : convert_revision : bd6352647798275a12d52d55a129cdddd8e25423
2006-04-06 15:00:11 -04:00
Gabe Black
a4b31e8f6b Enable register windows.
arch/alpha/isa_traits.hh:
arch/mips/isa_traits.cc:
    Turned the integer register file into a class instead of a typedef to an array.
arch/alpha/regfile.hh:
    Changed the integer register file into a class instead of a typedef to an array. Also put the parts of the register file, ie the int, float, and misc register files, pc, npc, and nnpc, behind accessor functions. Added a changeContext function, and ContextParam and ContextVal types, so that things like the register window can be changed through call backs.
arch/mips/isa_traits.hh:
    Turned the integer register file into a class instead of a typedef to an array. Also moved a "using namespace" into the namespace definition.
arch/sparc/isa_traits.hh:
    Turned the integer register file into a class instead of a typedef to an array. Also "fixed" the max number of src and dest regs. They may need to be even larger.
arch/sparc/regfile.hh:
    Changed the integer register file into a class instead of a typedef to an array. Also put the parts of the register file, ie the int, float, and misc register files, pc, npc, and nnpc, behind accessor functions. Added a changeContext function, and ContextParam and ContextVal types, so that things like the register window can be changed through call backs. Created setCWP and setAltGlobals functions for the IntRegFile.
cpu/cpu_exec_context.hh:
    Used the accessor functions for the register file, and added a changeRegFileContext function to call back into the RegFile. Used the RegFile clear function rather than memsetting it to 0.
cpu/exec_context.hh:
    Added the changeRegFileContext function.
cpu/exetrace.cc:
    Use the TheISA::NumIntRegs constant, and use readReg now that the integer register file is a class instead of an array.
cpu/exetrace.hh:
    Get the address of the regs object, now that it isn't an array.

--HG--
extra : convert_revision : ea2dd81be1c2e66b3c684af319eb58f8a77fd49c
2006-04-06 14:47:03 -04:00
Ali Saidi
bb80f71f21 fixes for new memory system
SConscript:
    comment out most devices
    add vport.cc
arch/alpha/arguments.cc:
arch/alpha/arguments.hh:
    push in alpha name space
    fix for new memory system
arch/alpha/faults.cc:
arch/alpha/faults.hh:
    Added an unimplemented fault that can be returned if a certain
    function isn't implemented
arch/alpha/freebsd/system.cc:
arch/alpha/linux/system.cc:
arch/alpha/stacktrace.cc:
arch/alpha/system.cc:
arch/alpha/tlb.hh:
arch/alpha/tru64/system.cc:
    fixed for new memory system
arch/alpha/tlb.cc:
    fixed for new memory system
    removed code that seems to have no purpose
arch/alpha/vtophys.cc:
arch/alpha/vtophys.hh:
    fixed for new memory system
    put in namespace AlphaISA
base/remote_gdb.cc:
    fix for new memory system
cpu/cpu_exec_context.cc:
cpu/cpu_exec_context.hh:
cpu/exec_context.hh:
    create two ports one of physical accesses and one for superpage accesses
    Add functions getVirtPort() getPhysPort() delVirtPort(). To get statically
    allocated physical or virtual ports or if an execcontext is passed in
    get a dynamically allocated virtual port
dev/alpha_console.cc:
dev/alpha_console.hh:
    Redo for new memory system
dev/io_device.cc:
dev/io_device.hh:
    new I/O devices for new memory system
kern/linux/events.cc:
kern/linux/printk.cc:
kern/linux/printk.hh:
kern/tru64/dump_mbuf.hh:
kern/tru64/printf.cc:
kern/tru64/printf.hh:
    Arguments now in namespaces
kern/tru64/tru64_events.cc:
mem/bus.cc:
    fix for new memory syste
mem/physical.hh:
    new addressranges function
    getPort should be public
mem/port.hh:
    Add write/read methods to functional port
    update getDeviceAddrRanges to have a list of both snoops and response lists
sim/pseudo_inst.cc:
sim/system.cc:
sim/system.hh:
    Update for new mem system
sim/vptr.hh:
    comment out code and replace with panics
    This will need to be fixed at some point, but it's not easy.

--HG--
extra : convert_revision : 41f41f422cfbab3751284d55cccb6ea64a7956e2
2006-04-06 00:51:46 -04:00
Ali Saidi
5936c79ba0 Add a functional port that is used to load the original binaries in FS
SE mode now has a port that goes to whatever toplevel mem object the
CPU sees that does the appropriate translation for syscall emulation

SConscript:
    translating port is a syscall emu only source
arch/alpha/system.cc:
base/loader/object_file.cc:
base/loader/object_file.hh:
    Use the new functional port to write the binaries into memory
cpu/cpu_exec_context.cc:
cpu/cpu_exec_context.hh:
cpu/simple/cpu.cc:
    We aren't always going to be writing straight to memory with syscalls
    support writing to a cache
mem/port.hh:
    Add a simple unidirectional functional port that panics on any incoming requests
mem/translating_port.hh:
    make translating port inherit from the simple port
sim/system.cc:
sim/system.hh:
    Add a functional port that is used to load the original binaries

--HG--
extra : convert_revision : 9096866d0b23e3aceea68394abb76e63c0f8fd8d
2006-03-30 18:06:00 -05:00
Ali Saidi
e196d20d9d Make TranslatingPort be a type of Port rather than something special
arch/alpha/arguments.cc:
    rather than returning 0, put a panic in... it will actually make us fix
    this rather than scratching our respective heads
base/loader/object_file.cc:
base/loader/object_file.hh:
    Object loader now takes a port rather than a translating port
cpu/cpu_exec_context.cc:
cpu/cpu_exec_context.hh:
sim/process.cc:
    Make translating port a type of port rather than anything special
cpu/simple/cpu.cc:
    no need to grab a port from the cpu anymore
mem/physical.cc:
    add an additional type of port to physicalmemory called "functional"
    Only used for functional accesses (loading binaries/syscall emu)
mem/port.hh:
    make readBlok/writeBlob virtual so translating port can do the
    translation first
mem/translating_port.cc:
mem/translating_port.hh:
    Make TranslatingPort inherit from Port
sim/system.cc:
    header file that doesn't exit removed

--HG--
extra : convert_revision : 89b08f6146bba61f5605678d736055feab2fe6f7
2006-03-30 15:59:49 -05:00
Kevin Lim
0b2deb2a88 Fixes for full system compiling.
arch/alpha/arguments.cc:
    There will not be a phys mem ptr in the XC in the newmem.  This read will have to go through something else.
arch/alpha/ev5.cc:
    Remove instantiations of these functions for the FastCPU, as the FastCPU is not really used.  Also this messed up the ability to specify which CPU models are being built.
cpu/exec_context.hh:
    Remove getPhysMemPtr() function.
cpu/exetrace.cc:
    Include sim/system.hh, and sort the includes.
cpu/simple/cpu.cc:
    Fixes for full system compilation.
kern/system_events.cc:
    Remove include of encumbered FullCPU.  The branch prediction will need to be fixed up in a more generic way in the future.

--HG--
extra : convert_revision : a8bbf562a277aa80e8f40112570c0a825298a05c
2006-03-30 10:42:55 -05:00
Gabe Black
fa763d2ecf Merge m5.eecs.umich.edu:/bk/newmem
into  ewok.(none):/home/gblack/m5/newmem

cpu/cpu_exec_context.cc:
    Hand merge

--HG--
rename : arch/alpha/registerfile.hh => arch/alpha/regfile.hh
extra : convert_revision : bd18966f7c37c67c2bc7ca2633b58f70ce64409c
2006-03-14 16:08:32 -05:00
Gabe Black
8e4ec55703 Changed the floating point register file into a class with appropriate accessor functions. The width of the floating point register to access can be specified, and if not, it will be accessed at its "natural" width. That is, the width of each individual register. Also, the functions which access the bit representation of floating point registers can use the blahblahBits functions now instead of blahblahInt.
arch/alpha/arguments.cc:
    Renamed readFloatRegInt to readFloatRegBits
arch/alpha/ev5.cc:
    Removed the Double from setFloatRegDouble
arch/alpha/registerfile.hh:
    Changed the floating point register file from a union of arrays to a class with appropriate accessor functions. The interface is necessary for SPARC.
arch/alpha/types.hh:
    Changed the FloatReg type from a union of uint64_t and double to a double, and defined a new type FloatRegBits which is a uint64_t and is used to return the bits which compose a floating point register rather than the value of the register.
arch/isa_parser.py:
    Adjusted the makeRead and makeWrite functions to generate the new versions of readFloatReg and setFloatReg.
base/remote_gdb.cc:
kern/tru64/tru64.hh:
    Replaced setFloatRegInt with setFloatRegBits
cpu/cpu_exec_context.cc:
    Removed the duplicated code for setting the floating point registers, and renamed the function to setFloatRegBits and readFloatRegBits.
cpu/cpu_exec_context.hh:
cpu/exec_context.hh:
cpu/o3/alpha_cpu_impl.hh:
cpu/o3/alpha_dyn_inst.hh:
cpu/o3/cpu.cc:
cpu/o3/cpu.hh:
cpu/o3/regfile.hh:
cpu/ozone/cpu.hh:
cpu/simple/cpu.hh:
    Implemented the new versions of the floating point read and set functions.
cpu/simple/cpu.cc:
    Replaced setFloatRegDouble with setFloatReg

--HG--
extra : convert_revision : 3dad06224723137f6033c335fb8f6395636767f2
2006-03-14 15:55:00 -05:00
Steve Reinhardt
e2b329d574 Replace Memory with MemObject; no need for two different levels of hierarchy there.
Get rid of addPort().
Change getPort() behavior on PhysicalMemory.

SConscript:
cpu/simple/cpu.hh:
sim/system.cc:
sim/system.hh:
    Replace Memory with MemObject.
cpu/base.hh:
    No need to declare Port here anymore.
cpu/cpu_exec_context.hh:
    Need PageTable definition.
cpu/simple/cpu.cc:
mem/physical.cc:
mem/physical.hh:
    Replace Memory with MemObject.
    Get rid of addPort(); allow getting anonymous ports with getPort().
mem/translating_port.hh:
    Remove unneeded header.
sim/process.cc:
    Replace Memory with MemObject.
    Change how initialization port gets set up to deal with change in
    addPort()/getPort().  Current solution is not ideal but it works.
sim/process.hh:
    Remove unneeded headers and declarations.
    Make LiveProcess::getDesc() abstract instead of panicing if called.
sim/syscall_emul.hh:
    Fix includes.

--HG--
extra : convert_revision : 11d4ffb54230038afcf7219cc46e51f809329a2f
2006-03-12 17:21:59 -05:00
Steve Reinhardt
37c860d334 Get rid of validInstAddr() & validDataAddr().
SE mode can now use page tables to determine which addresses are valid.

sim/process.cc:
sim/process.hh:
    Get rid of validInstAddr() & validDataAddr().
    SE mode can now use page tables to determine which addresses are valid.
    Also get rid of some Process object fields that were only used by those functions.

--HG--
extra : convert_revision : 74a25c0c2453bfc598eedacdbfccea1cf6493ba6
2006-03-12 16:01:41 -05:00
Gabe Black
2952ea7c82 Merge m5.eecs.umich.edu:/bk/newmem
into  ewok.(none):/home/gblack/m5/newmem

--HG--
extra : convert_revision : 58a5ae14fc8ac697206a3bfa1cf543a3579123d4
2006-03-10 17:01:58 -05:00
Ron Dreslinski
d07115f972 Add getMemPort to exec context proxy
--HG--
extra : convert_revision : a28c0410a63745b7455ad957c582c38319901cf0
2006-03-10 16:51:52 -05:00
Gabe Black
68d7382cf3 Eliminated TARGET_ALPHA, since THE_ISA provides the same function.
--HG--
extra : convert_revision : eb173a553b0782891e8b4a8e227bfb647390883a
2006-03-10 16:26:31 -05:00
Gabe Black
f102365bfe SimpleCPU compiles with merge.
arch/alpha/isa_traits.hh:
arch/alpha/linux/process.cc:
arch/alpha/process.cc:
arch/alpha/process.hh:
arch/alpha/tru64/process.cc:
base/chunk_generator.hh:
base/loader/elf_object.cc:
cpu/cpu_exec_context.cc:
cpu/cpu_exec_context.hh:
cpu/exec_context.hh:
cpu/simple/cpu.cc:
kern/linux/linux.hh:
kern/tru64/tru64.hh:
mem/packet.hh:
mem/page_table.cc:
mem/page_table.hh:
mem/physical.cc:
mem/request.hh:
mem/translating_port.cc:
sim/process.hh:
sim/system.cc:
    Fixing merged changes.

--HG--
extra : convert_revision : 2e94f21009395db654880fcb94ec806b6f5772c3
2006-03-09 19:21:35 -05:00
Gabe Black
872bbdfc33 Hand merge. Stuff probably doesn't compile.
--HG--
rename : arch/alpha/isa_desc => arch/alpha/isa/main.isa
rename : arch/alpha/alpha_linux_process.cc => arch/alpha/linux/process.cc
rename : arch/alpha/alpha_linux_process.hh => arch/alpha/linux/process.hh
rename : arch/alpha/alpha_tru64_process.cc => arch/alpha/tru64/process.cc
rename : arch/alpha/alpha_tru64_process.hh => arch/alpha/tru64/process.hh
rename : cpu/exec_context.cc => cpu/cpu_exec_context.cc
rename : cpu/exec_context.hh => cpu/cpu_exec_context.hh
extra : convert_revision : 7d1efcedd708815d985a951f6f010fbd83dc27e8
2006-03-09 18:35:28 -05:00
Ali Saidi
56cc760f6f Merge zizzer:/bk/multiarch
into  zeep.eecs.umich.edu:/z/saidi/work/m5.ma2

arch/alpha/isa_traits.hh:
arch/alpha/linux/process.cc:
arch/sparc/isa_traits.hh:
arch/sparc/linux/process.cc:
sim/process.cc:
    merge

--HG--
rename : cpu/exec_context.hh => cpu/cpu_exec_context.hh
extra : convert_revision : fea0155c8e23abbd0d5d5251abbd0f4d223fe935
2006-03-09 15:56:42 -05:00
Gabe Black
91545ac2bf Merge zizzer.eecs.umich.edu:/z/m5/Bitkeeper/m5
into  zizzer.eecs.umich.edu:/z/m5/Bitkeeper/multiarch

cpu/simple/cpu.cc:
    Hand Merge

--HG--
rename : cpu/exec_context.hh => cpu/cpu_exec_context.hh
extra : convert_revision : bf664b092f993d0f4675ce8e7df13645a920c1f4
2006-03-09 15:15:55 -05:00
Kevin Lim
7712232e55 Fixes to allow the ExecContext to be used for profiling.
cpu/base.cc:
    Change to be calls through the ExecContext instead of accessing the profile object directly.
cpu/cpu_exec_context.cc:
    Add functions to clear and sample the profile object.
cpu/cpu_exec_context.hh:
    Add functions to clear and sample the profile object.  These are not the most flexible functions; it might be better to eventually move the quiesce, profile, and store conditional stuff out of ExecContext so they don't clutter the interface.
cpu/exec_context.hh:
    Include functions to support using the profile object to clear itself and take samples.

--HG--
extra : convert_revision : 40849915fd51303673451515debb9ecdc7afb8c8
2006-03-07 22:21:39 -05:00
Kevin Lim
11aead894d Updates for the quiesceEvent that was added to the XC.
Also several files need to include system.hh or symtab.hh.  This is because exec_context.hh has less #includes than before, requiring some of the files that include it to include some other files as well.

arch/alpha/faults.cc:
    Avoid accessing XC directly.
arch/alpha/stacktrace.cc:
    StackTrace needs to include system.hh.
cpu/cpu_exec_context.cc:
    Update for change to CPUExecContext.
cpu/cpu_exec_context.hh:
    Make quiesce events use CPUExecContext instead of ExecContext.  Include functions to allow the quiesce event and last activate/suspend be accessed.
cpu/exec_context.hh:
    Include functions for quiesceEvent.
cpu/intr_control.cc:
    Needs to include cpu/exec_context.hh.
cpu/profile.cc:
    Needs to include symtab.hh for the symbol table.
cpu/profile.hh:
    Needs forward declare of ExecContext.
cpu/simple/cpu.cc:
    Rename xc to cpuXC.
dev/tsunami_cchip.cc:
    Needs to include exec_context.hh.
kern/kernel_stats.cc:
    Needs to include system.hh.
kern/linux/events.cc:
    Needs to include system.hh.

    Also avoid accessing objects directly from the XC.
kern/tru64/dump_mbuf.cc:
    Include symtab.hh for the SymbolTable and system.hh.
kern/tru64/tru64_events.cc:
    Include system.hh
sim/pseudo_inst.cc:
    Avoid accessing objects directly within the XC.

--HG--
extra : convert_revision : 78fe30d98cd20f7403fa216f772071458b675c84
2006-03-07 19:59:12 -05:00
Kevin Lim
20eced3ea0 Merge ktlim@zizzer:/bk/m5
into  zamp.eecs.umich.edu:/z/ktlim2/m5-proxyxc

Further changes still need to be made to the XC code.

arch/alpha/ev5.cc:
arch/alpha/freebsd/system.cc:
arch/alpha/linux/system.cc:
base/remote_gdb.cc:
cpu/cpu_exec_context.cc:
cpu/cpu_exec_context.hh:
cpu/simple/cpu.cc:
cpu/simple/cpu.hh:
kern/kernel_stats.cc:
sim/pseudo_inst.cc:
    Hand merge.

--HG--
rename : kern/freebsd/freebsd_system.cc => arch/alpha/freebsd/system.cc
rename : kern/linux/linux_system.cc => arch/alpha/linux/system.cc
rename : kern/linux/linux_threadinfo.hh => arch/alpha/linux/threadinfo.hh
rename : arch/alpha/alpha_linux_process.cc => arch/alpha/linux_process.cc
rename : arch/alpha/alpha_memory.cc => arch/alpha/tlb.cc
rename : arch/alpha/alpha_tru64_process.cc => arch/alpha/tru64_process.cc
rename : cpu/exec_context.cc => cpu/cpu_exec_context.cc
rename : cpu/exec_context.hh => cpu/cpu_exec_context.hh
extra : convert_revision : c1fe71fdd87d1fcd376f4feec69fc3fa29152e3e
2006-03-05 00:34:54 -05:00
Kevin Lim
f15e492375 Steps towards setting up the infrastructure to allow the new CPU model to work in full system mode.
The major change is renaming the old ExecContext to CPUExecContext, and creating two new classes, ExecContext (an abstract class), and ProxyExecContext (a templated class that derives from ExecContext).

Code outside of the CPU continues to use ExecContext as normal (other than not being able to access variables within the XC).  The CPU uses the CPUExecContext, or however else it stores its own state.  It then creates a ProxyExecContext, templated on the class used to hold its state.  This proxy is passed to any code outside of the CPU that needs to access the XC.  This allows code outside of the CPU to use the ExecContext interface to access any state needed, without knowledge of how that state is laid out.

Note that these changes will not compile without the accompanying revision to automatically rename the shadow registers.

SConscript:
    Include new file, cpu_exec_context.cc.
arch/alpha/alpha_linux_process.cc:
arch/alpha/alpha_memory.cc:
arch/alpha/alpha_tru64_process.cc:
arch/alpha/arguments.cc:
arch/alpha/isa/decoder.isa:
arch/alpha/stacktrace.cc:
arch/alpha/vtophys.cc:
base/remote_gdb.cc:
cpu/intr_control.cc:
    Avoid directly accessing objects within the XC.
arch/alpha/ev5.cc:
    Avoid directly accessing objects within the XC.

    KernelStats have been moved to the BaseCPU instead of the XC.
arch/alpha/isa_traits.hh:
    Remove clearIprs().  It wasn't used very often and it did not work well with the proxy ExecContext.
cpu/base.cc:
    Place kernel stats within the BaseCPU instead of the ExecContext.

    For now comment out the profiling code sampling until its exact location is decided upon.
cpu/base.hh:
    Kernel stats are now in the BaseCPU instead of the ExecContext.
cpu/base_dyn_inst.cc:
cpu/base_dyn_inst.hh:
cpu/memtest/memtest.cc:
cpu/memtest/memtest.hh:
    Changes to support rename of old ExecContext to CPUExecContext.  See changeset for more details.
cpu/exetrace.cc:
    Remove unneeded include of exec_context.hh.
cpu/intr_control.hh:
cpu/o3/alpha_cpu_builder.cc:
    Remove unneeded include of exec_context.hh
cpu/o3/alpha_cpu.hh:
cpu/o3/alpha_cpu_impl.hh:
cpu/o3/cpu.cc:
cpu/o3/cpu.hh:
cpu/simple/cpu.cc:
cpu/simple/cpu.hh:
    Changes to support rename of old ExecContext to CPUExecContext.  See changeset for more details.

    Also avoid accessing anything directly from the XC.
cpu/pc_event.cc:
    Avoid accessing objects directly from the XC.
dev/tsunami_cchip.cc:
    Avoid accessing objects directly within the XC>
kern/freebsd/freebsd_system.cc:
kern/linux/linux_system.cc:
kern/linux/linux_threadinfo.hh:
kern/tru64/dump_mbuf.cc:
kern/tru64/tru64.hh:
kern/tru64/tru64_events.cc:
sim/syscall_emul.cc:
sim/syscall_emul.hh:
    Avoid accessing objects directly within the XC.
kern/kernel_stats.cc:
kern/kernel_stats.hh:
    Kernel stats no longer exist within the XC.
kern/system_events.cc:
    Avoid accessing objects directly within the XC.  Also kernel stats are now in the BaseCPU.
sim/process.cc:
sim/process.hh:
    Avoid accessing regs directly within an ExecContext.  Instead use a CPUExecContext to initialize the registers and copy them over.
cpu/cpu_exec_context.cc:
    Rename old ExecContext to CPUExecContext.  This is used by the old CPU models to store any necessary architectural state.  Also include the ProxyExecContext, which is used to access the CPUExecContext's state in code outside of the CPU.
cpu/cpu_exec_context.hh:
    Rename old ExecContext to CPUExecContext.  This is used by the old CPU models to store any necessary architectural state.  Also include the ProxyExecContext, which is used to access the CPUExecContext's state in code outside of the CPU.

    Remove kernel stats from the ExecContext.
sim/pseudo_inst.cc:
    Kernel stats now live within the CPU.

    Avoid accessing objects directly within the XC.

--HG--
rename : cpu/exec_context.cc => cpu/cpu_exec_context.cc
rename : cpu/exec_context.hh => cpu/cpu_exec_context.hh
extra : convert_revision : a75393a8945c80cca225b5e9d9c22a16609efb85
2006-03-04 15:18:40 -05:00