8353b1e21f
Make the bus is occupied for none broadcast packets as well.
Gabe Black
2006-10-10 23:28:33 -04:00
477a3b0b61
Merge zizzer:/n/wexford/x/gblack/m5/newmem_bus into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-10 22:52:52 -04:00
1de8eae43a
Debugging info
Ron Dreslinski
2006-10-10 22:50:36 -04:00
59dd317cb5
Put in an accounting mechanism and an assert to make sure something doesn't try to send another packet while it's still waiting for the bus.
Gabe Black
2006-10-10 22:10:08 -04:00
404b2a951d
Fixed a corner case and simplified the logic in Packet::intersect.
Gabe Black
2006-10-10 17:49:31 -04:00
27c59dc370
Merge zizzer:/n/wexford/x/gblack/m5/newmem_bus into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-10 17:32:24 -04:00
aff3d92c00
Some more code cleanup
Ron Dreslinski
2006-10-10 17:25:50 -04:00
549412b333
Changed the bus to use a bool to keep track of retries rather than a pointer
Gabe Black
2006-10-10 17:24:03 -04:00
5f9aca531d
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-10 17:18:09 -04:00
995146ead7
Fix some more mem leaks, still some left Update retry mechanism
Ron Dreslinski
2006-10-10 17:10:56 -04:00
012556ecf9
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-10 15:56:18 -04:00
9e008d73d5
Fix cshr Retry's Fix Upgrades being blocked by slave
Ron Dreslinski
2006-10-10 15:53:25 -04:00
3a9eb598c3
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-10 15:04:55 -04:00
72bf1c011e
Merge ktlim@zizzer:/bk/newmem into zamp.eecs.umich.edu:/z/ktlim2/clean/o3-merge/newmem
Kevin Lim
2006-10-10 11:04:21 -04:00
f9284b1111
Updates refs.
Kevin Lim
2006-10-10 11:04:05 -04:00
fe8b912c03
Merge zizzer:/z/m5/Bitkeeper/newmem into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-10 02:36:04 -04:00
3fa5e4b6b8
Yet another fix to the HasData command attribute.
Ron Dreslinski
2006-10-10 02:33:30 -04:00
b40798070b
Actually set the HasData attribute on Read Responses
Ron Dreslinski
2006-10-10 02:21:03 -04:00
89e80ccc20
Fix another merge issue
Ron Dreslinski
2006-10-10 02:00:37 -04:00
a0472af008
Merge zizzer:/z/m5/Bitkeeper/newmem into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-10 01:57:57 -04:00
e5b13138b1
Two minor fixes.
Kevin Lim
2006-10-10 01:49:46 -04:00
cc78d86661
Fix several bugs pertaining to upgrades/mem leaks.
Ron Dreslinski
2006-10-10 01:32:18 -04:00
5582e60966
Fixed a bug where a packet was attempted to be sent even though another packet was waiting for the bus.
Gabe Black
2006-10-10 00:49:27 -04:00
ab44417282
Fixes to the bus, and added fields to the packet.
Gabe Black
2006-10-09 23:24:21 -04:00
bdde892d66
Merge ktlim@zizzer:/bk/newmem into zamp.eecs.umich.edu:/z/ktlim2/clean/o3-merge/newmem
Kevin Lim
2006-10-09 22:59:56 -04:00
a9ae6c8656
Comment out code that messed up SMT (but will be needed eventually).
Kevin Lim
2006-10-09 22:49:58 -04:00
ec8a437b2c
Handle NACK's that occur from devices on the same bus. Not fully implemented yet, but good enough for single level cache coherence
Ron Dreslinski
2006-10-09 20:18:00 -04:00
5448517da4
updated reference output
Gabe Black
2006-10-09 19:55:49 -04:00
843888c489
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-09 19:35:53 -04:00
9356bcda7b
Fix a typo preventing compilation
Ron Dreslinski
2006-10-09 19:20:28 -04:00
e03b9c9939
Fix how upgrades work. Remove some dead code.
Ron Dreslinski
2006-10-09 19:15:24 -04:00
92bf23bed6
Be sure to delete packet and sender state if the cache is blocked.
Kevin Lim
2006-10-09 19:14:14 -04:00
af7315c7dc
Fix caches plus sampling switch over.
Kevin Lim
2006-10-09 19:13:06 -04:00
13ac9a419d
One step closet to having NACK's work.
Ron Dreslinski
2006-10-09 18:52:20 -04:00
a23c6a7193
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-09 18:19:35 -04:00
187dcb18bf
Potentially functional partially timed bandwidth limitted bus model.
Gabe Black
2006-10-09 18:12:45 -04:00
727dea78c4
Update configs for cpu_id
Ron Dreslinski
2006-10-09 17:31:58 -04:00
c4dba7a8ed
Fix a typo in the printf
Ron Dreslinski
2006-10-09 17:30:54 -04:00
094c6de663
Multiprogrammed workload, need to generate ref's for it yet. But Nate wanted the config. Not sure on the naming convention for tests.
Ron Dreslinski
2006-10-09 17:25:43 -04:00
fd27c229b6
Fix a bitwise operation that was accidentally a logical operation.
Ron Dreslinski
2006-10-09 17:18:34 -04:00
b9fb4d4870
Make memtest work with 8 memtesters
Ron Dreslinski
2006-10-09 17:13:50 -04:00
d7c1557e7e
Merge zizzer:/z/m5/Bitkeeper/newmem into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-09 16:48:58 -04:00
45732376f6
Add more DPRINTF's fix a supply condition.
Ron Dreslinski
2006-10-09 16:47:55 -04:00
afce51d10a
Set size properly on uncache accesses Don't use the senderState after you get a succesful sendTiming. Not guarnteed to be correct
Ron Dreslinski
2006-10-09 16:37:02 -04:00
d95b23b81f
Fix outstanding bug (FS#158).
Kevin Lim
2006-10-09 11:01:19 -04:00
6a2d6c0f83
Fix checker bug.
Kevin Lim
2006-10-09 11:00:31 -04:00
bc732b59fd
Have cpus send snoop ranges
Ron Dreslinski
2006-10-09 01:04:37 -04:00
095d5991f5
Put a check in so people know not to create more than 8 memtesters.
Ron Dreslinski
2006-10-09 00:31:24 -04:00
95ca4f9d4a
Merge zizzer:/z/m5/Bitkeeper/newmem into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-09 00:28:26 -04:00
0087061681
Don't create a response if one isn't needed.
Ron Dreslinski
2006-10-09 00:27:41 -04:00
4f93c43d34
Don't block responses even if the cache is blocked.
Ron Dreslinski
2006-10-09 00:27:03 -04:00
6c7ab02682
Update the Memtester, commit a config file/test for it.
Ron Dreslinski
2006-10-09 00:26:10 -04:00
67a114fc29
add in checkpoint restoration option, you can restore a checkpoint by giving a directory, and then giving a checkpoint number, the earliest checkpoint is 1, the latest is N. the default checkpoint directory is the cwd.
Lisa Hsu
2006-10-09 00:12:16 -04:00
4167c3c026
Update memory assertion to check for whole range.
Kevin Lim
2006-10-09 00:09:44 -04:00
67c20dd9dd
Merge zizzer:/bk/newmem into zed.eecs.umich.edu:/z/hsul/work/m5/newmem
Lisa Hsu
2006-10-08 23:19:03 -04:00
97c1f6eff7
post checkpoint restoration the bus ranges need to be re-initialized for ALL pci devs, not just ide.
Lisa Hsu
2006-10-08 23:18:19 -04:00
d52117d1e3
add in serialization of AtomicSimpleCPU _status. This is needed because right now unserializing breaks an assert since CPU status is not saved. Kev says that this will break uniform serialization across CPUs since each type of CPU has its own "status" enum set. So, the repercussions are that if you serialize in this CPU, you must first unserialize in this CPU before switching to something else you want.
Lisa Hsu
2006-10-08 23:16:40 -04:00
a2bdddcc6f
Merge zizzer.eecs.umich.edu:/bk/newmem into vm1.(none):/home/stever/bk/newmem-head
Steve Reinhardt
2006-10-08 19:11:19 -07:00
91c76278b9
Set cpu_id params (required by ll/sc code now).
Steve Reinhardt
2006-10-08 19:11:06 -07:00
ce6c752ede
update for m5 base linux. (the last changes were for the latest m5hack, i.e. with nate's stuff in it).
Lisa Hsu
2006-10-08 22:05:34 -04:00
31f3f24214
Fixes for Port proxies and proxy parameters.
Steve Reinhardt
2006-10-08 18:26:59 -07:00
0a3e4d56e5
Update stats for functional path fix
Ron Dreslinski
2006-10-08 21:08:27 -04:00
4cfddc0d77
Make sure to propogate sendFunctional calls with functional not atomic.
Ron Dreslinski
2006-10-08 20:47:50 -04:00
5cb1840b31
Fixes for functional path.
Ron Dreslinski
2006-10-08 20:30:42 -04:00
2df9053bb0
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-08 19:14:09 -04:00
e65f0cef3c
Only respond if the pkt needs a response. Fix an issue with memory handling writebacks.
Ron Dreslinski
2006-10-08 19:05:48 -04:00
8a539a774f
Merge zizzer:/z/m5/Bitkeeper/newmem into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-08 18:49:30 -04:00
1345183a89
Move away from using the statusChange function on snoops. Clean up snooping code in general.
Ron Dreslinski
2006-10-08 18:48:03 -04:00
0c574f1069
missing else
Gabe Black
2006-10-08 18:45:21 -04:00
a82f017591
bus changes
Gabe Black
2006-10-08 18:44:49 -04:00
5df93cc1cd
Replace tests of LOCKED/UNCACHEABLE flags with isLocked()/isUncacheable().
Steve Reinhardt
2006-10-08 14:48:24 -07:00
911381321b
Update ref stats: ll/sc, cpu_id, new kernel (?)
Steve Reinhardt
2006-10-08 17:07:23 -04:00
63023ba4c2
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-08 14:09:13 -04:00
00481d1f19
A possible implementation of a multiplexed bus.
Gabe Black
2006-10-08 14:08:58 -04:00
34b697cd04
Add in HasData, and move the define of NUM_MEM_CMDS to a more visible location.
Gabe Black
2006-10-08 14:04:49 -04:00
d3fba5aa30
Implement Alpha LL/SC support for SimpleCPU (Atomic & Timing) and PhysicalMemory. *No* support for caches or O3CPU. Note that properly setting cpu_id on all CPUs is now required for correct operation.
Steve Reinhardt
2006-10-08 10:53:24 -07:00
be36c808f7
Rename some vars for clarity.
Steve Reinhardt
2006-10-08 10:43:31 -07:00
c2f954ac69
Allocate new thread stacks and shared mem region via Process page table for Tru64 thread library emulation.
Steve Reinhardt
2006-10-08 04:29:40 -04:00
8949d813ff
Clean up configs.
Kevin Lim
2006-10-08 01:12:42 -04:00
b17421da20
Record numCycles properly.
Kevin Lim
2006-10-08 00:55:05 -04:00
d48ea81ba2
Updates to O3 CPU. It should now work in FS mode, although sampling still has a bug.
Kevin Lim
2006-10-08 00:53:41 -04:00
c0e53b6d4c
Merge ktlim@zizzer:/bk/newmem into zamp.eecs.umich.edu:/z/ktlim2/clean/o3-merge/newmem
Kevin Lim
2006-10-07 23:35:32 -04:00
abc27ae691
Merge zizzer:/bk/newmem into zeep.pool:/z/saidi/work/m5.newmem.head
Ali Saidi
2006-10-07 14:49:10 -04:00
3afc69df77
Merge ktlim@zizzer:/bk/newmem into zizzer.eecs.umich.edu:/.automount/zamp/z/ktlim2/clean/o3-merge/newmem
Kevin Lim
2006-10-07 13:41:49 -04:00
fe762278e7
Updates to bring MemTest closer to working with newmem. Ron still needs to do the initial setup and configuration for it to work properly.
Kevin Lim
2006-10-07 13:37:22 -04:00
413a5379b8
Update stats for change in functional path in cache
Ron Dreslinski
2006-10-07 12:58:37 -04:00
467c17fbd9
Fix a missing pointer
Ron Dreslinski
2006-10-07 12:55:37 -04:00
fdaed2c7ae
No need to keep trying to request the data bus if we are already waiting.
Ron Dreslinski
2006-10-07 12:20:29 -04:00
df3014a726
Add mechanism for caches to handle failure of the fast path on responses.
Ron Dreslinski
2006-10-07 12:02:59 -04:00
2c336f0a1f
Merge zizzer:/z/m5/Bitkeeper/newmem into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-07 11:37:18 -04:00
178d114fa5
Fix infinite writebacks bug in cache.
Ron Dreslinski
2006-10-07 11:36:55 -04:00
984579a6ad
Update refs.
Kevin Lim
2006-10-07 11:32:10 -04:00
c60c61b19e
Merge zizzer:/bk/newmem into zeep.pool:/z/saidi/work/m5.newmem.head
Ali Saidi
2006-10-06 21:46:04 -04:00
b7832555d5
system.cc: Make new_page() check for an out of memory condition
Ali Saidi
2006-10-06 21:45:34 -04:00
862825f997
Merge zizzer.eecs.umich.edu:/bk/newmem into zeep.eecs.umich.edu:/home/gblack/m5/newmem_bus
Gabe Black
2006-10-06 17:10:13 -04:00
c42a7bc4f6
Merge zizzer:/z/m5/Bitkeeper/newmem into zazzer.eecs.umich.edu:/z/rdreslin/m5bk/newmemcleanest
Ron Dreslinski
2006-10-06 09:28:16 -04:00
dfdb683fb9
Another thread number removed
Ron Dreslinski
2006-10-06 09:27:59 -04:00
1b6653b6f7
Remove threadnum from cache everywhere for now Fix so that blocking for the same reason doesn't fail. I.E. multiple writebacks want to set the blocked flag.
Ron Dreslinski
2006-10-06 09:15:53 -04:00
48e89a9d1e
Merge zizzer.eecs.umich.edu:/z/m5/Bitkeeper/newmem into zizzer.eecs.umich.edu:/.automount/zooks/y/ksewell/research/m5-sim/m5-clean
Korey Sewell
2006-10-06 04:24:02 -04:00
84dac99e2e
add SMT hello world test - 2 threads
Korey Sewell
2006-10-06 04:23:27 -04:00
Gabe Black