gem5.git
17 years agoBetter handling of scons -u targets.
Steve Reinhardt [Mon, 4 Dec 2006 17:09:36 +0000 (09:09 -0800)]
Better handling of scons -u targets.

--HG--
extra : convert_revision : 7bf0688a1c83d8385b77a59a1c75040e9624c0ae

17 years agoSConstruct:
Steve Reinhardt [Mon, 4 Dec 2006 16:59:53 +0000 (08:59 -0800)]
SConstruct:
Couple minor fixes.

SConstruct:
    Couple minor fixes.

--HG--
extra : convert_revision : 25f3c12570287334c2cbd1cf9b8227043a57e7d1

17 years agoimport os.path.join as joinpath
Steve Reinhardt [Mon, 4 Dec 2006 16:55:06 +0000 (08:55 -0800)]
import os.path.join as joinpath

--HG--
extra : convert_revision : 200612675e49908b9ff9c965aede35a657241391

17 years agoMerge vm1.(none):/home/stever/bk/newmem-head
Steve Reinhardt [Sun, 3 Dec 2006 06:26:40 +0000 (22:26 -0800)]
Merge vm1.(none):/home/stever/bk/newmem-head
into  vm1.(none):/home/stever/bk/newmem-cache2

--HG--
extra : convert_revision : 321f7fcc8bd6c6aaaab92d10172814f4d07d5e65

17 years agoSupport better param conversions to/from numeric subclasses.
Steve Reinhardt [Sun, 3 Dec 2006 06:24:52 +0000 (22:24 -0800)]
Support better param conversions to/from numeric subclasses.

--HG--
extra : convert_revision : 2ccb75b0912a384789458710fd9bbb65626f839e

17 years agoFix help strings on GenRepl params.
Steve Reinhardt [Sun, 3 Dec 2006 06:23:46 +0000 (22:23 -0800)]
Fix help strings on GenRepl params.

--HG--
extra : convert_revision : 520814e193b9e86b6410f3ab98d62ed131d295aa

17 years agoMake cache compression policy a runtime virtual thing
Steve Reinhardt [Sun, 3 Dec 2006 06:22:58 +0000 (22:22 -0800)]
Make cache compression policy a runtime virtual thing
instead of a template policy.

--HG--
extra : convert_revision : 6a4ac7a189a950390a973fdfce94f56190de92db

17 years agoMerge zizzer.eecs.umich.edu:/bk/newmem
Steve Reinhardt [Sun, 3 Dec 2006 06:12:26 +0000 (22:12 -0800)]
Merge zizzer.eecs.umich.edu:/bk/newmem
into  vm1.(none):/home/stever/bk/newmem-head

--HG--
extra : convert_revision : 4a077b463b938c54b546b00e586d8609c24ae465

17 years agoDelete src/oldmem.
Steve Reinhardt [Sun, 3 Dec 2006 06:11:24 +0000 (01:11 -0500)]
Delete src/oldmem.

util/make_release.py:
    src/oldmem gone from repo, no need to delete here.

--HG--
extra : convert_revision : 570fa1b8d7144376cf13a010160a39d1c1cccbc2

17 years agoMerge zizzer.eecs.umich.edu:/bk/newmem
Steve Reinhardt [Sun, 3 Dec 2006 03:52:26 +0000 (19:52 -0800)]
Merge zizzer.eecs.umich.edu:/bk/newmem
into  vm1.(none):/home/stever/bk/newmem-head

--HG--
extra : convert_revision : 4c8de6e4c6f729c83e92abd81ea6c1347e647756

17 years agoFixes for MIPS_SE compiling. Regressions seem to work, but Korey should make sure...
Kevin Lim [Sat, 2 Dec 2006 18:33:46 +0000 (13:33 -0500)]
Fixes for MIPS_SE compiling.  Regressions seem to work, but Korey should make sure these changes (commit especially) work okay.

src/cpu/o3/commit_impl.hh:
src/cpu/o3/fetch_impl.hh:
    Fixes for MIPS_SE compile.

--HG--
extra : convert_revision : fde9616f8e72b397c5ca965774172372cff53790

17 years agostats update
Lisa Hsu [Sat, 2 Dec 2006 16:11:58 +0000 (11:11 -0500)]
stats update

--HG--
extra : convert_revision : b47d3817d204a43e0afb89aafc8dacf619c3f910

17 years agodon't blow away the whole destination directory
Nathan Binkert [Sat, 2 Dec 2006 06:33:18 +0000 (22:33 -0800)]
don't blow away the whole destination directory

--HG--
extra : convert_revision : 7370bad15cc30e75ebb0c8685324d8db06fc2936

17 years agochange this to be a quick one so that it's in the regressions every night - it's...
Lisa Hsu [Fri, 1 Dec 2006 18:51:49 +0000 (13:51 -0500)]
change this to be a quick one so that it's in the regressions every night - it's only maybe 15 min. long.

tests/configs/twosys-tsunami-simple-atomic.py:
    don't need this import

--HG--
rename : tests/long/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/config.ini => tests/quick/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/config.ini
rename : tests/long/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/config.out => tests/quick/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/config.out
rename : tests/long/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/console.drivesys.sim_console => tests/quick/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/console.drivesys.sim_console
rename : tests/long/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/console.testsys.sim_console => tests/quick/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/console.testsys.sim_console
rename : tests/long/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/m5stats.txt => tests/quick/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/m5stats.txt
rename : tests/long/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/stderr => tests/quick/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/stderr
rename : tests/long/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/stdout => tests/quick/80.netperf-stream/ref/alpha/linux/twosys-tsunami-simple-atomic/stdout
rename : tests/long/80.netperf-stream/test.py => tests/quick/80.netperf-stream/test.py
extra : convert_revision : 68497b2ef8b21590cb6c636485703e46dc616513

17 years agoMerge zizzer:/bk/newmem
Lisa Hsu [Fri, 1 Dec 2006 06:24:16 +0000 (01:24 -0500)]
Merge zizzer:/bk/newmem
into  zed.eecs.umich.edu:/z/hsul/work/m5/newmem

--HG--
extra : convert_revision : b62ca8009105aad7173bdbc5d528de243d21a82c

17 years agoadd a simple netperf-stream test to the long tests.
Lisa Hsu [Fri, 1 Dec 2006 06:24:01 +0000 (01:24 -0500)]
add a simple netperf-stream test to the long tests.

tests/SConscript:
    add a new configuration for two-system tests (atomic simple only)

--HG--
extra : convert_revision : 16c260ab16f38779fe17b1cab18f36d5c7a70846

17 years agoMerge zizzer.eecs.umich.edu:/bk/newmem
Nathan Binkert [Fri, 1 Dec 2006 04:53:30 +0000 (20:53 -0800)]
Merge zizzer.eecs.umich.edu:/bk/newmem
into  iceaxe.:/Volumes/work/m5/incoming

--HG--
extra : convert_revision : 05060a06e0b6d66a9c1e7005c233047c6e8ba15f

17 years agoGet rid of the old release-edits script and create make_release.py
Nathan Binkert [Fri, 1 Dec 2006 04:50:47 +0000 (20:50 -0800)]
Get rid of the old release-edits script and create make_release.py
which takes care of almost everything needed for putting together
a release.

--HG--
extra : convert_revision : b05d418a1002633b1286591eb8a8588ba33f5df1

17 years agoUpdate stats to match writeback fix that was made
Ron Dreslinski [Thu, 30 Nov 2006 20:01:49 +0000 (15:01 -0500)]
Update stats to match writeback fix that was made

--HG--
extra : convert_revision : 3e0ed2b374d8d96798ea9b3416c9e5579cafacda

17 years agonetperf-maerts-client.rcS:
Lisa Hsu [Thu, 30 Nov 2006 16:53:33 +0000 (11:53 -0500)]
netperf-maerts-client.rcS:
change /netperf/netperf to /netperf-bin/netperf
nat-netperf-maerts-client.rcS:
bad comment that went with the file - accidentally committed but probably doesn't matter, i ust eliminated an ivlb in the script.

configs/boot/nat-netperf-maerts-client.rcS:
    replace netperf/netperf with netperf-bin/netperf
configs/boot/netperf-maerts-client.rcS:
    change /netperf/netperf to /netperf-bin/netperf

--HG--
extra : convert_revision : 32fed0042e267f315d3e688ebc4b66d7002b85f0

17 years agocscope-find.py:
Steve Reinhardt [Wed, 29 Nov 2006 21:17:41 +0000 (13:17 -0800)]
cscope-find.py:
Write directly to 'cscope.files' and run 'cscope -b' .
Now this script does everything automatically.
cscope-index.py:
Rename: util/cscope-find.py -> util/cscope-index.py

util/cscope-find.py:
    Write directly to 'cscope.files' and run 'cscope -b' .
    Now this script does everything automatically.

--HG--
rename : util/cscope-find.py => util/cscope-index.py
extra : convert_revision : cd6fa5cc0c2146f7184c9213956aff67c7cb9341

17 years agoMerge ktlim@zizzer:/bk/newmem
Kevin Lim [Wed, 29 Nov 2006 21:08:19 +0000 (16:08 -0500)]
Merge ktlim@zizzer:/bk/newmem
into  zamp.eecs.umich.edu:/z/ktlim2/clean/tmp/test-regress

--HG--
extra : convert_revision : 3142f68356458ecd2677c30e9cf0a65005b782c2

17 years agoChange the connecting of the physPort and virtPort to the memory object below the...
Kevin Lim [Wed, 29 Nov 2006 21:07:55 +0000 (16:07 -0500)]
Change the connecting of the physPort and virtPort to the memory object below the CPU to happen every time activateContext is called.  The overhead is probably a little higher than necessary, but allows these connections to properly be made when there are CPUs that are inactive until they are switched in.

Right now this introduces a minor memory leak as old physPorts and virtPorts are not deleted when new ones are created.  A flyspray task has been created for this issue.  It can not be resolved until we determine how the bus will handle giving out ID's to functional ports that may be deleted.

src/cpu/o3/cpu.cc:
src/cpu/simple/atomic.cc:
src/cpu/simple/timing.cc:
    Change the setup of the physPort and virtPort to instead happen every time the CPU has a context activated.  This is a little high overhead, but keeps it working correctly when the CPU does not have a physical memory attached to it until it switches in (like the case of switch CPUs).
src/cpu/o3/thread_context.hh:
    Change function from being called at init() to just being called whenever the memory ports need to be connected.
src/cpu/o3/thread_context_impl.hh:
    Update this to not delete the port if it's the same as the virtPort.
src/cpu/thread_context.hh:
    Change function from being called at init() to whenever the memory ports need to be connected.
src/cpu/thread_state.cc:
    Instead of initializing the ports, simply connect them, deleting any old ports that might exist.  This allows these functions to be called multiple times.
src/cpu/thread_state.hh:
    Ports are no longer initialized, but rather connected at context activation time.

--HG--
extra : convert_revision : e399ce5dfbd6ad658c953a7c9c7b69b89a70219e

17 years agoAdd in O3CPU to default CPU list.
Kevin Lim [Wed, 29 Nov 2006 16:50:03 +0000 (11:50 -0500)]
Add in O3CPU to default CPU list.

--HG--
extra : convert_revision : 4aaaae058cb763580ea0b9019d4a9346938121d4

17 years agoadd 2.0b2 release notes
Ali Saidi [Tue, 28 Nov 2006 21:02:13 +0000 (16:02 -0500)]
add 2.0b2 release notes

--HG--
extra : convert_revision : ce34f8086f682cc732bf868f6b9700e42c604ca3

17 years agoMerge ktlim@zizzer:/bk/newmem
Kevin Lim [Tue, 28 Nov 2006 16:41:17 +0000 (11:41 -0500)]
Merge ktlim@zizzer:/bk/newmem
into  zamp.eecs.umich.edu:/z/ktlim2/clean/tmp/test-regress

--HG--
extra : convert_revision : ffc7931d7da153b421b3c838a0968e484fd182ec

17 years agoRemove assertion. It's not needed and messes up writebacks when a 2 level cache...
Kevin Lim [Tue, 28 Nov 2006 16:41:08 +0000 (11:41 -0500)]
Remove assertion.  It's not needed and messes up writebacks when a 2 level cache is used in a uniprocessor setting.

--HG--
extra : convert_revision : 020a9799cd7177fdb85a767701d6fcb8cf018827

17 years agoAdd TRACING_ON setting for m5.prof.
Steve Reinhardt [Mon, 27 Nov 2006 07:16:24 +0000 (02:16 -0500)]
Add TRACING_ON setting for m5.prof.

--HG--
extra : convert_revision : ebda49bff30d76d3209acce55458d3f4e29594d3

17 years agoInclude check for making sure caches are enabled.
Kevin Lim [Sun, 26 Nov 2006 16:46:58 +0000 (11:46 -0500)]
Include check for making sure caches are enabled.

--HG--
extra : convert_revision : e3902b065db524ebe5bf762e44a840133ccb8d75

17 years agoAdd no-op versions of ivlb and ivle back in for backwards compatibility.
Steve Reinhardt [Fri, 24 Nov 2006 17:32:33 +0000 (12:32 -0500)]
Add no-op versions of ivlb and ivle back in for backwards compatibility.

--HG--
extra : convert_revision : 383b72c130b20f3d7cde4e08fa36a481f3c0bf7c

17 years agoMerge zizzer.eecs.umich.edu:/z/m5/Bitkeeper/newmem
Steve Reinhardt [Thu, 23 Nov 2006 15:46:24 +0000 (10:46 -0500)]
Merge zizzer.eecs.umich.edu:/z/m5/Bitkeeper/newmem
into  zizzer.eecs.umich.edu:/z/stever/bk/newmem-head

--HG--
extra : convert_revision : 154bc605c62b1e51c32e65916d4c2eda3a3f22fd

17 years agoDo a functional access to levels above on a read as a temporary solution for L2's...
Ron Dreslinski [Thu, 23 Nov 2006 01:20:38 +0000 (20:20 -0500)]
Do a functional access to levels above on a read as a temporary solution for L2's in FS

Fix a small writeback bug when missing in the L2 in atomic mode

src/mem/bus.cc:
    Fix a comment to make sense
src/mem/cache/cache_impl.hh:
    Do a functional access to levels above on a read as a temporary solution for L2's in FS
    Also fix a small writeback miss in L2 issue
src/mem/cache/coherence/simple_coherence.hh:
src/mem/cache/coherence/uni_coherence.cc:
src/mem/cache/coherence/uni_coherence.hh:
    Do a functional access to levels above on a read as a temporary solution for L2's in FS
tests/quick/00.hello/ref/alpha/linux/o3-timing/m5stats.txt:
tests/quick/00.hello/ref/alpha/linux/simple-timing/m5stats.txt:
tests/quick/01.hello-2T-smt/ref/alpha/linux/o3-timing/m5stats.txt:
    Update ref's for writeback changes

--HG--
extra : convert_revision : 937febd577b16b7fd97a5a68acaf53541828a251

17 years agoFix typo.
Kevin Lim [Mon, 20 Nov 2006 16:44:27 +0000 (11:44 -0500)]
Fix typo.

--HG--
extra : convert_revision : 2dd830c6b3b5df894608b7596250b0181a3dfdf0

17 years agoMerge zizzer.eecs.umich.edu:/z/m5/Bitkeeper/newmem
Steve Reinhardt [Sun, 19 Nov 2006 23:38:12 +0000 (18:38 -0500)]
Merge zizzer.eecs.umich.edu:/z/m5/Bitkeeper/newmem
into  zizzer.eecs.umich.edu:/z/stever/bk/newmem-head

--HG--
extra : convert_revision : c2b7784377d85df5b8ee39c891cd3da9907410d8

17 years agoUpdate Virtual and Physical ports.
Kevin Lim [Sun, 19 Nov 2006 22:43:03 +0000 (17:43 -0500)]
Update Virtual and Physical ports.

src/cpu/o3/alpha/cpu_impl.hh:
    Handle the PhysicalPort and VirtualPort in the ThreadState.
src/cpu/o3/cpu.cc:
    Initialize the thread context.
src/cpu/o3/thread_context.hh:
    Add new function to initialize thread context.
src/cpu/o3/thread_context_impl.hh:
    Use code now put into function.
src/cpu/simple_thread.cc:
    Move code to ThreadState and use the new helper function.
src/cpu/simple_thread.hh:
    Remove init() in this derived class; use init() from ThreadState base class.
src/cpu/thread_state.cc:
    Move setting up of Physical and Virtual ports here.  Change getMemFuncPort() to connectToMemFunc(), which connects a port to a functional port of the memory object below the CPU.
src/cpu/thread_state.hh:
    Update functions.

--HG--
extra : convert_revision : ff254715ef0b259dc80d08f13543b63e4024ca8d

17 years agoMerge zizzer:/bk/newmem
Ron Dreslinski [Sat, 18 Nov 2006 03:01:18 +0000 (22:01 -0500)]
Merge zizzer:/bk/newmem
into  zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest

--HG--
extra : convert_revision : 1fc55d7d5707bb7c63790aab306ca5ea8ade5fab

17 years agoMake an initialization pass for the thread context and set the [phys,virt]Port correctly
Ron Dreslinski [Sat, 18 Nov 2006 02:55:28 +0000 (21:55 -0500)]
Make an initialization pass for the thread context and set the [phys,virt]Port correctly

src/cpu/simple/atomic.cc:
src/cpu/simple/timing.cc:
    Call the thread context initialization

--HG--
extra : convert_revision : d7dc2a8b893dc670077b7f6150d4b710a1778620

17 years agoadd warn_once which will print any given warning message
Nathan Binkert [Thu, 16 Nov 2006 21:18:21 +0000 (13:18 -0800)]
add warn_once which will print any given warning message
only once.

--HG--
extra : convert_revision : b64bb495c1bd0c4beb3db6ca28fad5af4d05ef8e

17 years agoImplement a single config file to encompass all of the SPEC
Nathan Binkert [Thu, 16 Nov 2006 21:10:38 +0000 (13:10 -0800)]
Implement a single config file to encompass all of the SPEC
CPU2000 stuff, and use it in all of the tests that currently
use SPEC

--HG--
extra : convert_revision : 8cd26a597e51a90b6d2810d344a075f5aa0f011b

17 years agoimplement RUSAGE_CHILDREN for getrusage since it's trivial
Nathan Binkert [Thu, 16 Nov 2006 21:08:29 +0000 (13:08 -0800)]
implement RUSAGE_CHILDREN for getrusage since it's trivial

--HG--
extra : convert_revision : bc12b3b2e9ee02f42c437cbc20680ea00e19a801

17 years agoImplement current working directory for LiveProcesses
Nathan Binkert [Thu, 16 Nov 2006 20:43:11 +0000 (12:43 -0800)]
Implement current working directory for LiveProcesses

--HG--
extra : convert_revision : a2d3cf29ab65c61af27d82a8c421a41a19fd5aeb

18 years agoAdd L2 cache option to fs.py --l2cache
Ron Dreslinski [Wed, 15 Nov 2006 23:22:15 +0000 (18:22 -0500)]
Add L2 cache option to fs.py --l2cache

--HG--
extra : convert_revision : 5bdd1129c3b23e91d441e7b83f6a824ef7740fab

18 years agoMerge zizzer:/bk/newmem
Ron Dreslinski [Tue, 14 Nov 2006 23:41:37 +0000 (18:41 -0500)]
Merge zizzer:/bk/newmem
into  zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest

--HG--
extra : convert_revision : 8d61b474428d494b1a5382e4cf95934ad54e35dd

18 years agoVarious fixes to delete packet and request a little better.
Kevin Lim [Tue, 14 Nov 2006 22:22:32 +0000 (17:22 -0500)]
Various fixes to delete packet and request a little better.

src/cpu/simple/timing.cc:
    Various updates for deleting requests more properly.

    The major change is moving the deletion of the fetch request/packet to after the instruction has executed and completed.  This should fix a few bugs because Ron's memory system didn't expect a call for a functional access while a timing access was being processed.

--HG--
extra : convert_revision : c7cf114bb1ff3cdaa7b0a40ed4c5302dc9d3a522

18 years agoMerge zizzer:/bk/newmem
Ron Dreslinski [Tue, 14 Nov 2006 22:19:57 +0000 (17:19 -0500)]
Merge zizzer:/bk/newmem
into  zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest

--HG--
extra : convert_revision : b216fcdb2632dce68ac18932b0c13408eb1aeaf4

18 years agoFix bugs around uni-coherence invalidates being propogated properly.
Ron Dreslinski [Tue, 14 Nov 2006 22:15:05 +0000 (17:15 -0500)]
Fix bugs around uni-coherence invalidates being propogated properly.

src/mem/bus.cc:
    Make it so that invalidates being sent from the responder up don't call the responder
    but they should also not Panic.
src/mem/packet.hh:
    If we don't have data in the packet, don't call deleteData:
    Example: InvalidateRequests never have data.

--HG--
extra : convert_revision : 18766bc9f3bb4d852ac651d094254d347abd1634

18 years agoMerge zizzer:/bk/newmem
Lisa Hsu [Tue, 14 Nov 2006 18:00:05 +0000 (13:00 -0500)]
Merge zizzer:/bk/newmem
into  zed.eecs.umich.edu:/z/hsul/work/sparc/m5

--HG--
extra : convert_revision : 6abd919711966eaaa157483557a3f953b02dde01

18 years agointerrupts.hh:
Lisa Hsu [Tue, 14 Nov 2006 17:59:57 +0000 (12:59 -0500)]
interrupts.hh:
make a likewise updateIntrInfo for Sparc that's blank so it doesn't fart on build

src/arch/sparc/interrupts.hh:
    make a likewise updateIntrInfo for Sparc that's blank so it doesn't fart on build

--HG--
extra : convert_revision : 5f469d0cf897479b42703104cd801a8ef923fcae

18 years agoIf all the targets aren't satisfied, reinitialize the packet.
Ron Dreslinski [Tue, 14 Nov 2006 15:09:13 +0000 (10:09 -0500)]
If all the targets aren't satisfied, reinitialize the packet.

--HG--
extra : convert_revision : 5b0a977a162a1b881b97a3185fb386cc76632a4a

18 years agoUpdate atomic and functional paths for snoops as well
Ron Dreslinski [Tue, 14 Nov 2006 06:38:42 +0000 (01:38 -0500)]
Update atomic and functional paths for snoops as well

--HG--
extra : convert_revision : 566d73438efb87ca683e4dee23454d880db3dfc7

18 years agoUpdate phase param in the .py file for the cpus
Ron Dreslinski [Tue, 14 Nov 2006 06:13:26 +0000 (01:13 -0500)]
Update phase param in the .py file for the cpus

--HG--
extra : convert_revision : cd2eb8c00adcb34b8693a4d1a66187927c0f6803

18 years agoUpdate bus bridges now that snoop ranges are passed properly
Ron Dreslinski [Tue, 14 Nov 2006 06:12:52 +0000 (01:12 -0500)]
Update bus bridges now that snoop ranges are passed properly

src/mem/bridge.cc:
    Update brdiges, now that snoop addresses are properly forwarded.
    Bus bridge should only handle snoops on the second phase (SNOOP_COMMIT)
src/mem/bus.cc:
src/mem/bus.hh:
    Make sure if a busBridge has access to both things that snoop and things that respond it only takes the request once

--HG--
extra : convert_revision : 26cc9ee4429be45d4476fa435e0e9a54843c2509

18 years agoMake cpu's capable of having a phase shift
Ron Dreslinski [Tue, 14 Nov 2006 06:10:36 +0000 (01:10 -0500)]
Make cpu's capable of having a phase shift

--HG--
extra : convert_revision : 7f082ba5c1cd2445aec731950c31a877aac23a75

18 years agoFix a bug to handle the fact that a CPU can send Functional accesses while a sendTimi...
Ron Dreslinski [Tue, 14 Nov 2006 03:37:22 +0000 (22:37 -0500)]
Fix a bug to handle the fact that a CPU can send Functional accesses while a sendTiming has not returned in the call stack.

src/mem/cache/base_cache.cc:
    Sometimes a functional access comes while waiting on a outstanding packet being sent.
    This could be because Timing CPU does some post processing on the recvTiming which send functional access.
    Either the CPU should leave the pkt/req around (so They can be referenced in the mem system). Or the mem
    system should remove them from outstanding lists and reinsert them if they fail in the sendTiming.

    I did the later, eventually we should consider doing the former if that is the correct behavior.

--HG--
extra : convert_revision : be41e0d2632369dca9d7c15e96e5576d7583fe6a

18 years agoMerge zizzer:/bk/newmem
Ron Dreslinski [Tue, 14 Nov 2006 02:59:50 +0000 (21:59 -0500)]
Merge zizzer:/bk/newmem
into  zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest

--HG--
extra : convert_revision : 253766a17bb5e109f8ad76b3c54e443db5274ef5

18 years agoIf we didn't satisfy all targets, reset the packet we are requesting with.
Ron Dreslinski [Tue, 14 Nov 2006 02:34:25 +0000 (21:34 -0500)]
If we didn't satisfy all targets, reset the packet we are requesting with.

--HG--
extra : convert_revision : 736372131b046eccf3520292fb3c086dc568d918

18 years agoFix some errors related to snooping and functional access in the bus
Ron Dreslinski [Tue, 14 Nov 2006 02:33:01 +0000 (21:33 -0500)]
Fix some errors related to snooping and functional access in the bus

src/mem/bus.cc:
    Only call snoop once per port, need to fix it so snoop ranges that overlap aren't added to list
    Functional accesses that call snoop and it goes to a higher bus may change the src, reset it after each snoop.

--HG--
extra : convert_revision : 7276059c798a85cb9d138ccc5531298ecd055c13

18 years agoFix problems with snoop ranges not working properly on functional accesses
Ron Dreslinski [Tue, 14 Nov 2006 00:56:34 +0000 (19:56 -0500)]
Fix problems with snoop ranges not working properly on functional accesses

src/mem/bus.cc:
    Actually return the snoop list when asked for it.
    Don't get stuck in infinite functional loops

--HG--
extra : convert_revision : 8e6dafbd10b30d48d28b6b5d4b464e8e8f6a3ddc

18 years agoChanges needed for a bus from CPU->L1
Ron Dreslinski [Tue, 14 Nov 2006 00:12:45 +0000 (19:12 -0500)]
Changes needed for a bus from CPU->L1

src/cpu/simple/atomic.cc:
    Make the atomic cpu return 0 on snoops.

--HG--
extra : convert_revision : aad96ad36e0c764c7cfef8b0c8e97877574f5845

18 years agoSince cpus now send out snoop ranges, remove it from the cache.
Ron Dreslinski [Tue, 14 Nov 2006 00:00:50 +0000 (19:00 -0500)]
Since cpus now send out snoop ranges, remove it from the cache.

--HG--
extra : convert_revision : 82882eb131aa66eba9f281b64db21d5cbfefb1b9

18 years agoMake CPU models signal to update the snoop ranges
Ron Dreslinski [Mon, 13 Nov 2006 23:51:16 +0000 (18:51 -0500)]
Make CPU models signal to update the snoop ranges

--HG--
extra : convert_revision : 717b62510f28a69af99453309fbbb458359eeb2a

18 years agomain.i:
Lisa Hsu [Mon, 13 Nov 2006 21:18:47 +0000 (16:18 -0500)]
main.i:
swig still thought the default arg for simulate() was -1, make it MaxTick

--HG--
extra : convert_revision : 2bcbef7e5e5d28cf55645fdc53d43e3953b1a11c

18 years agoUpdate splash2 config files
Ron Dreslinski [Mon, 13 Nov 2006 21:09:47 +0000 (16:09 -0500)]
Update splash2 config files

configs/splash2/run.py:
    Fix MaxTick for splash configs
configs/splash2/cluster.py:
    Add a config that allows clusters of CPU's to be attached to a single L1

--HG--
extra : convert_revision : 1bb0a0c5f4889316940a9858be90ae2eaa849f1a

18 years agoExpose debugBreakCycle through swig and get rid of
Nathan Binkert [Mon, 13 Nov 2006 20:20:08 +0000 (12:20 -0800)]
Expose debugBreakCycle through swig and get rid of
the Debug param context

--HG--
extra : convert_revision : 40e9dcfa9faedbe0c90a43f908f20a7c14ded6a4

18 years agoClean up comment
Nathan Binkert [Mon, 13 Nov 2006 20:18:36 +0000 (12:18 -0800)]
Clean up comment

--HG--
extra : convert_revision : 75c6415c49305e2f8cf7f833c66a46a4dee503ea

18 years agoParse CC/CXX early so that we use the correct compiler for testing dependencies/versi...
Ali Saidi [Mon, 13 Nov 2006 18:26:36 +0000 (13:26 -0500)]
Parse CC/CXX early so that we use the correct compiler for testing dependencies/versions/libraries/etc

--HG--
extra : convert_revision : b34ee03e44b35df30b746c4c570316db75d6ffe5

18 years agoMerge ktlim@zizzer:/bk/newmem
Kevin Lim [Mon, 13 Nov 2006 07:49:13 +0000 (02:49 -0500)]
Merge ktlim@zizzer:/bk/newmem
into  zamp.eecs.umich.edu:/z/ktlim2/clean/newmem-busfix

--HG--
extra : convert_revision : 6ef2249bfa3f7149830efdb42a313422090da7d7

18 years agoMore interrupt reworking.
Kevin Lim [Mon, 13 Nov 2006 07:49:03 +0000 (02:49 -0500)]
More interrupt reworking.

--HG--
extra : convert_revision : 40dfbb72c4e418c54e909c54dad5fe6ef7017cb4

18 years agoMerge zizzer.eecs.umich.edu:/bk/newmem
Steve Reinhardt [Mon, 13 Nov 2006 06:04:00 +0000 (22:04 -0800)]
Merge zizzer.eecs.umich.edu:/bk/newmem
into  vm1.(none):/home/stever/bk/newmem-head

--HG--
extra : convert_revision : faab7569deefde94c20133b2f70a8567bcaa2960

18 years agoMake setRegWithEffect do something in SE mode.
Steve Reinhardt [Mon, 13 Nov 2006 06:03:42 +0000 (22:03 -0800)]
Make setRegWithEffect do something in SE mode.

--HG--
extra : convert_revision : 88fdaa403fe6d083f8c8fc064cb0d0d6a8b8daf8

18 years agoChange warn to DPRINTF.
Kevin Lim [Mon, 13 Nov 2006 05:26:38 +0000 (00:26 -0500)]
Change warn to DPRINTF.

--HG--
extra : convert_revision : 746bdf92334d220158eb0eb6bf113b4dcedbb354

18 years agoUpdate output refs. Some FS statistics will change (namely the ITB) due to the recen...
Kevin Lim [Mon, 13 Nov 2006 05:24:22 +0000 (00:24 -0500)]
Update output refs.  Some FS statistics will change (namely the ITB) due to the recent TLB changes.  Now PAL mode accesses are counted as hits in the TLB.

tests/quick/00.hello/ref/alpha/linux/o3-timing/config.ini:
tests/quick/00.hello/ref/alpha/linux/o3-timing/config.out:
tests/quick/00.hello/ref/alpha/linux/o3-timing/stderr:
tests/quick/00.hello/ref/alpha/linux/o3-timing/stdout:
tests/quick/00.hello/ref/alpha/linux/simple-atomic/config.ini:
tests/quick/00.hello/ref/alpha/linux/simple-atomic/config.out:
tests/quick/00.hello/ref/alpha/linux/simple-timing/config.ini:
tests/quick/00.hello/ref/alpha/linux/simple-timing/config.out:
tests/quick/00.hello/ref/alpha/tru64/o3-timing/config.ini:
tests/quick/00.hello/ref/alpha/tru64/o3-timing/config.out:
tests/quick/00.hello/ref/alpha/tru64/o3-timing/stderr:
tests/quick/00.hello/ref/alpha/tru64/o3-timing/stdout:
tests/quick/00.hello/ref/alpha/tru64/simple-atomic/config.ini:
tests/quick/00.hello/ref/alpha/tru64/simple-atomic/config.out:
tests/quick/00.hello/ref/alpha/tru64/simple-timing/config.ini:
tests/quick/00.hello/ref/alpha/tru64/simple-timing/config.out:
tests/quick/01.hello-2T-smt/ref/alpha/linux/o3-timing/config.ini:
tests/quick/01.hello-2T-smt/ref/alpha/linux/o3-timing/config.out:
tests/quick/01.hello-2T-smt/ref/alpha/linux/o3-timing/m5stats.txt:
tests/quick/01.hello-2T-smt/ref/alpha/linux/o3-timing/stderr:
tests/quick/01.hello-2T-smt/ref/alpha/linux/o3-timing/stdout:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-atomic-dual/config.ini:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-atomic-dual/config.out:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-atomic-dual/m5stats.txt:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-atomic/config.ini:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-atomic/config.out:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-atomic/m5stats.txt:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-timing-dual/config.ini:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-timing-dual/config.out:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-timing-dual/m5stats.txt:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-timing/config.ini:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-timing/config.out:
tests/quick/10.linux-boot/ref/alpha/linux/tsunami-simple-timing/m5stats.txt:
tests/quick/20.eio-short/ref/alpha/eio/simple-atomic/config.ini:
tests/quick/20.eio-short/ref/alpha/eio/simple-atomic/config.out:
tests/quick/20.eio-short/ref/alpha/eio/simple-timing/config.ini:
tests/quick/20.eio-short/ref/alpha/eio/simple-timing/config.out:
    Update refs.

--HG--
extra : convert_revision : 6798c5753d4d7bd7b5667d59cf564012b781ce8a

18 years agoFix typo.
Kevin Lim [Mon, 13 Nov 2006 04:31:29 +0000 (23:31 -0500)]
Fix typo.

--HG--
extra : convert_revision : 05db10e20d33302fe830d5759b8881b1233aca87

18 years agoFix for regression failure.
Kevin Lim [Mon, 13 Nov 2006 04:30:09 +0000 (23:30 -0500)]
Fix for regression failure.

src/cpu/o3/fetch_impl.hh:
    Fetch needs to make sure it isn't waiting on an Icache access.

--HG--
extra : convert_revision : b53eb58b9e5a00bdb394134586d1f84f84d1c6e1

18 years agoMerge ktlim@zamp:./local/clean/tmp/test-regress
Kevin Lim [Mon, 13 Nov 2006 02:57:58 +0000 (21:57 -0500)]
Merge ktlim@zamp:./local/clean/tmp/test-regress
into  zamp.eecs.umich.edu:/z/ktlim2/clean/newmem-busfix

--HG--
extra : convert_revision : b98236507bb8996ce605b48b5a5a6a7aac297dc5

18 years agoFix for non-FS compile.
Kevin Lim [Mon, 13 Nov 2006 02:49:51 +0000 (21:49 -0500)]
Fix for non-FS compile.

--HG--
extra : convert_revision : 661b412b0ae670181b89cb7dbc5e9d813804aa7a

18 years agoCreate a module called internal where swigged stuff goes.
Nathan Binkert [Mon, 13 Nov 2006 02:49:16 +0000 (18:49 -0800)]
Create a module called internal where swigged stuff goes.
Rename cc_main to internal.main

--HG--
extra : convert_revision : e938005f600fbf8a43435e29426a948f4501f072

18 years agoUpdates to support new interrupt processing and removal of PcPAL.
Kevin Lim [Mon, 13 Nov 2006 01:15:30 +0000 (20:15 -0500)]
Updates to support new interrupt processing and removal of PcPAL.

src/arch/alpha/interrupts.hh:
    No need for this now that the ThreadContext is being used to set these IPRs in interrupts.
    Also split up the interrupt checking from the updating of the IPL and interrupt summary.
src/arch/alpha/tlb.cc:
    Check the PC for whether or not it's in PAL mode, not the addr.
src/cpu/o3/alpha/cpu.hh:
    Split up getting the interrupt from actually processing the interrupt.
src/cpu/o3/alpha/cpu_impl.hh:
    Splut up the processing of interrupts.
src/cpu/o3/commit_impl.hh:
    Update for ISA-oriented interrupt changes.
src/cpu/o3/fetch_impl.hh:
    Fix broken if statement from PcPAL updates, and properly populate the request fields.

    Also more debugging output.
src/cpu/ozone/cpu_impl.hh:
    Updates for ISA-oriented interrupt stuff.
src/cpu/ozone/front_end_impl.hh:
    Populate request fields properly.
src/cpu/simple/base.cc:
    Update for interrupt stuff.

--HG--
extra : convert_revision : 9bac3f9ffed4948ee788699b2fa8419bc1ca647c

18 years agoUpdate for maxtick in splash2/memtest configs
Ron Dreslinski [Sun, 12 Nov 2006 16:42:07 +0000 (11:42 -0500)]
Update for maxtick in splash2/memtest configs

configs/example/memtest.py:
configs/splash2/run.py:
    Update for maxtick

--HG--
extra : convert_revision : 94106625be1ebc2b614db16720a4861e47222c0b

18 years agoUpdate refs for functional access fixes
Ron Dreslinski [Sun, 12 Nov 2006 16:07:34 +0000 (11:07 -0500)]
Update refs for functional access fixes

--HG--
extra : convert_revision : aa34e0e7254daf20ecb7c14d430f08927a8fb9ca

18 years agoMerge zizzer:/bk/newmem
Ron Dreslinski [Sun, 12 Nov 2006 15:41:18 +0000 (10:41 -0500)]
Merge zizzer:/bk/newmem
into  zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest

--HG--
extra : convert_revision : 498304c24435437f8f1942bb8aeafe69ba22a089

18 years agoPhysical memory overrides the tport version of recvFunctional, need to do the
Ron Dreslinski [Sun, 12 Nov 2006 14:30:12 +0000 (09:30 -0500)]
Physical memory overrides the tport version of recvFunctional, need to do the
check here for responses that match as well

--HG--
extra : convert_revision : 69c3628a381a9da885fab0272abf40c3411a5f0f

18 years agoHandle packets being deleted by lower level properly.
Ron Dreslinski [Sun, 12 Nov 2006 14:06:15 +0000 (09:06 -0500)]
Handle packets being deleted by lower level properly.
Fixes for Mem Leak associated with Writebacks.

src/mem/cache/miss/mshr_queue.cc:
    Fixes for Mem Leak associated with Writebacks. (Double Delete removed)

--HG--
extra : convert_revision : 7a52ddd57da35995896f2c4438a58aa53f762416

18 years agoDon't insert reponses into the list more than once
Ron Dreslinski [Sun, 12 Nov 2006 12:16:34 +0000 (07:16 -0500)]
Don't insert reponses into the list more than once
If you get inserted in the front, reschedule the event

--HG--
extra : convert_revision : eccbacf5ec85600e5b68eb554fee2c0e2b65e965

18 years agoMove code before a early return to make sure it is executed on all paths
Ron Dreslinski [Sun, 12 Nov 2006 11:44:05 +0000 (06:44 -0500)]
Move code before a early return to make sure it is executed on all paths

--HG--
extra : convert_revision : cfdd5b6911422fbb733677c43d027aa4407fbc85

18 years agoYet another small bug in mem system related to flow control
Ron Dreslinski [Sun, 12 Nov 2006 11:36:33 +0000 (06:36 -0500)]
Yet another small bug in mem system related to flow control

src/mem/cache/cache_impl.hh:
    When upgrades change to readEx make sure to allocate the block
    Fix dprintf

--HG--
extra : convert_revision : 8700a7e47ad042c8708302620b907849c4bfdded

18 years agoFix functional access errors related to delayed respnoses in cachePort
Ron Dreslinski [Sun, 12 Nov 2006 11:35:39 +0000 (06:35 -0500)]
Fix functional access errors related to delayed respnoses in cachePort

src/mem/cache/base_cache.cc:
    On a delayed response, be sure to call the fixPacket wrapper to toggle hasData flag.
src/mem/packet.cc:
src/mem/packet.hh:
    Create a wrapper to toggle the hasData flag on delayed responses

--HG--
extra : convert_revision : 1ced8d4e3dc12a059fb7636d59e429cd3dd46901

18 years agoFix Typo
Nathan Binkert [Sun, 12 Nov 2006 07:57:20 +0000 (23:57 -0800)]
Fix Typo

--HG--
extra : convert_revision : 4f5b610f364876b29ad0e04f1757e4b42d1f2bd8

18 years agoset TRACING_ON one way or another explicitly in the
Nathan Binkert [Sun, 12 Nov 2006 04:46:56 +0000 (20:46 -0800)]
set TRACING_ON one way or another explicitly in the
SConscript file instead of basing it on DEBUG

--HG--
extra : convert_revision : 6e6807cc4350ef92baeaaabfeb3dc0bb785128ba

18 years agoDelete the actual PseudoInst file
Nathan Binkert [Sun, 12 Nov 2006 01:30:37 +0000 (17:30 -0800)]
Delete the actual PseudoInst file

--HG--
extra : convert_revision : 54e3dfbc1022fcc663a9c0e67f6434395d1fe33c

18 years agoGet rid of the ParamContext for pseudo instructions and move
Nathan Binkert [Sun, 12 Nov 2006 01:22:10 +0000 (17:22 -0800)]
Get rid of the ParamContext for pseudo instructions and move
the parameters to the BaseCPU object.

--HG--
extra : convert_revision : 557292cffb40918133647b0c9ac653ee5112df2e

18 years agoThe Lock_Flag_DepTag went away earlier, and using TheISA gives the false impression...
Gabe Black [Sat, 11 Nov 2006 12:16:24 +0000 (07:16 -0500)]
The Lock_Flag_DepTag went away earlier, and using TheISA gives the false impression that this code is ISA independent.

--HG--
extra : convert_revision : 67d9e51702efbe5f5244268e3753328a6cf1a1d5

18 years agoCertain header files should only be used in FS.
Gabe Black [Sat, 11 Nov 2006 12:15:16 +0000 (07:15 -0500)]
Certain header files should only be used in FS.

src/arch/alpha/faults.hh:
    Only use pagetable.hh in FS
src/arch/alpha/pagetable.hh:
    pagetable.hh should only be included in FS, so protecting it internally should be unnecessary.
src/cpu/exetrace.cc:
    Only use tlb.hh in FS

--HG--
extra : convert_revision : 91ea61f2e7970e7146b6d407ee250fcb20cd4d48

18 years agoMore fixes for functional accesses. It now makes the writeback memory leak to crash...
Ron Dreslinski [Sat, 11 Nov 2006 03:45:50 +0000 (22:45 -0500)]
More fixes for functional accesses.  It now makes the writeback memory leak to crash all configs.
Working on that now.

src/mem/cache/base_cache.cc:
    Keep a list of the responders so we can search them on functional accesses.
src/mem/cache/base_cache.hh:
    Properly put things on a list for responses so we can search the list.
    Also, be sure to check the outgoing ports lists on a functional access (factor some common code out there)
src/mem/cache/cache_impl.hh:
    Properly return when the first read hit on a functional access.
    Make sure to call to check the other ports list of packets before forwarding it out.

--HG--
extra : convert_revision : 1d21cb55ff29c15716617efc48441329707c088a

18 years agoBig fix for functional access, where we forgot to copy the last byte on write interse...
Ron Dreslinski [Sat, 11 Nov 2006 03:41:21 +0000 (22:41 -0500)]
Big fix for functional access, where we forgot to copy the last byte on write intersections.

src/mem/packet.cc:
    Make sure to copy the whole data (we were one byte short)
src/mem/tport.cc:
    Fix for the proper semantics of fixPacket

--HG--
extra : convert_revision : 215e05db9099d427afd4994f5b29079354c847d8

18 years agoMerge zizzer:/bk/newmem
Ali Saidi [Sat, 11 Nov 2006 01:17:54 +0000 (20:17 -0500)]
Merge zizzer:/bk/newmem
into  zeep.pool:/z/saidi/work/m5.newmem

--HG--
extra : convert_revision : e4f9bb663099662a94c5522e6b4955c2a83bac8d

18 years agofix endian issues with condition codes
Ali Saidi [Sat, 11 Nov 2006 01:17:42 +0000 (20:17 -0500)]
fix endian issues with condition codes
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

18 years agoMerge ktlim@zizzer:/bk/newmem
Kevin Lim [Fri, 10 Nov 2006 20:35:06 +0000 (15:35 -0500)]
Merge ktlim@zizzer:/bk/newmem
into  zamp.eecs.umich.edu:/z/ktlim2/clean/o3-merge/newmem

--HG--
extra : convert_revision : 35e2ff6ce62281299ad98dca64ba04a3a8a6757c

18 years agoElaborated on the tlb stubs so that they just set the physical address to the virtual...
Gabe Black [Fri, 10 Nov 2006 20:32:15 +0000 (15:32 -0500)]
Elaborated on the tlb stubs so that they just set the physical address to the virtual address.

--HG--
extra : convert_revision : 41478abc4d21d504420f6842338675c0767f7cf9

18 years agoFixed up DepTags a little. I think NumMicroIntRegs shouldn't be added to Ctrl_Base_De...
Gabe Black [Fri, 10 Nov 2006 20:30:59 +0000 (15:30 -0500)]
Fixed up DepTags a little. I think NumMicroIntRegs shouldn't be added to Ctrl_Base_DepTag.

--HG--
extra : convert_revision : 2ebb3eb781441ba936c8d8bb1f42e4c0840aff2e