Andrew Waterman
5828e6042e
Work around https://github.com/ucb-bar/firrtl/issues/299
2016-09-14 11:47:10 -07:00
Andrew Waterman
c3ddff809b
Move PRCI from Coreplex to always-on block, where it belongs
2016-09-14 11:01:05 -07:00
Andrew Waterman
5566bf1b13
Don't route PLIC interrupts through PRCI
...
The PLIC is local to the Coreplex, and PRCI should not be.
2016-09-14 11:01:05 -07:00
mwachs5
47acbf928b
Give AsyncCrossing slave interfaces registers visibility into when they were written ( #288 )
2016-09-14 00:17:26 -07:00
Howard Mao
bdb7b1de36
move tilelink-agnostic counters from uncore to util package
2016-09-13 20:47:05 -07:00
Howard Mao
1882241493
move junctions utils into top-level utils package
2016-09-13 20:47:04 -07:00
Henry Cook
7dd4492abb
First cut at refactoring unittests into a top-level utility. Individual tests co-located with their DUT. No functional changes.
2016-09-13 20:30:14 -07:00
Wesley W. Terpstra
d23ab7370d
tilelink2: Unit Test for the RegisterCrossing
2016-09-13 18:33:56 -07:00
Wesley W. Terpstra
cc88bf1b08
junctions: give unit tests more time
2016-09-13 18:33:56 -07:00
Wesley W. Terpstra
acedd3688a
tilelink2: unit test for the clock crossing
2016-09-13 18:33:56 -07:00
Wesley W. Terpstra
c8e6d47884
tilelink2: add a clock crossing adapter
2016-09-13 18:33:56 -07:00
Wesley W. Terpstra
44501cdbf8
crossings: change defaults to sync=3 for safer settling time
...
Make the matching AsyncQueue depth=8 to support full throughput
2016-09-13 18:33:56 -07:00
Wesley W. Terpstra
3348236320
junctions: remove obsolete Handshaker crossing
2016-09-13 18:33:56 -07:00
Wesley W. Terpstra
fe6a67dd0e
tilelink2: add a RegisterCrossing primitive
2016-09-13 18:33:53 -07:00
Wesley W. Terpstra
d75f9d6a34
junctions: add an AsyncQueue
2016-09-13 17:38:18 -07:00
Wesley W. Terpstra
8142406d2e
junctions: refactor the Crossing type
2016-09-13 15:51:18 -07:00
Wesley W. Terpstra
ecdfb528c5
crossing: refactor AsyncDecoupled to provide AsyncDecoupledCrossing with no clock domain
2016-09-13 15:51:18 -07:00
Wesley W. Terpstra
33a05786db
tilelink2 RAMModel: fix put, get, putAck, getAck case ( #282 )
...
This case should result in undefined data for the Get.
It was previously requiring the Get to return the new Put data,
which is only guaranteed by a FIFO device.
2016-09-13 15:44:36 -07:00
Henry Cook
632b5896b9
Delete TestGraphs.scala
...
Re-do later using Fuzzer
2016-09-13 13:29:48 -07:00
Henry Cook
e318c29d48
[tilelink2] Fuzzer: Allow noise-making to be parameterized. Better comments.
2016-09-13 12:25:57 -07:00
Henry Cook
05100c12a7
Merge branch 'master' of github.com:ucb-bar/rocket-chip into monitor
2016-09-13 11:18:18 -07:00
Andrew Waterman
61cbe6164d
Add option to execute JAL from decode stage
...
This is particularly helpful for designs that don't have a BTB, but
it becomes the critical path for designs with RVC. Caveat emptor.
2016-09-13 02:32:00 -07:00
Wesley W. Terpstra
606f19a17f
tilelink2: RegisterRouter Unit Test
2016-09-12 22:13:39 -07:00
Wesley W. Terpstra
7005422651
tilelink2 HintHandler: don't HintAck in the middle of a multibeat op
2016-09-12 19:06:35 -07:00
Wesley W. Terpstra
273d3a73f2
tilelink2: Unit Test passes!
2016-09-12 18:39:50 -07:00
Wesley W. Terpstra
9874bc553a
tilelink2: Fragmenter supports Hints
2016-09-12 17:31:59 -07:00
Wesley W. Terpstra
42955a0490
tilelink2: HintHandler optimize to nothing if unneeded
2016-09-12 17:31:16 -07:00
Wesley W. Terpstra
94761f714d
tilelink2 HintHandler: fill in correct sink in responses
2016-09-12 17:26:40 -07:00
Wesley W. Terpstra
ca5f98f138
tilelink2: Hints are not special
...
Hints have a TransferSize limit just like all other message types.
2016-09-12 17:15:28 -07:00
Henry Cook
ad8e563c89
[tilelink2] Fuzzer: Rewrite of fuzzer
...
Multiple bug-fixes and actual source id generation.
2016-09-12 17:00:58 -07:00
Henry Cook
0b0c891179
[tilelink2] Monitor: Allow zero-mask PutPartials
...
this will require a larger address refactoring TBD
2016-09-12 17:00:50 -07:00
Henry Cook
c57b52ec86
tilelink2 Fragmenter: bugfix using D.hasData
2016-09-12 16:58:21 -07:00
Henry Cook
82681179cb
[tilelink2] Edges: add size to addr_lo.
...
addr_lo cannot correctly be deciphered from the mask alone.
OxC still has addr_lo === 0 if size is >1.
2016-09-12 16:58:09 -07:00
Andrew Waterman
88440ebf89
Use PseudoLRU in BTB when possible (for powers of two)
2016-09-12 16:52:03 -07:00
Andrew Waterman
266a2f24bd
Disable Mul early out by default if XLen == 32
...
With a default unroll of 8, it doesn't help performance, but costs area.
2016-09-12 16:50:08 -07:00
Andrew Waterman
96185e4b16
tighten an assert condition
...
dcache.s1_kill is a don't-care if dcache.req.valid wasn't previously high
2016-09-12 16:49:46 -07:00
Andrew Waterman
beb141a20b
Allow M, A, D, C extensions to be disabled in misa register
2016-09-12 16:49:46 -07:00
Howard Mao
f3cdeb08c6
pass nMemChannels to coreplex through CoreplexConfig
2016-09-12 12:40:10 -07:00
Howard Mao
9d9f90646d
allow configuration of simulation memory latency
2016-09-12 12:33:50 -07:00
Henry Cook
a21b04a7c1
playground for making different DAGs to use as DUTs
2016-09-12 10:32:45 -07:00
Henry Cook
0671d5d637
Initial version of fuzzer and simple ram fuzz test
2016-09-12 10:32:45 -07:00
Wesley W. Terpstra
7760459b76
tilelink2 RegisterRouter: add RegField test patterns
2016-09-12 10:32:25 -07:00
Wesley W. Terpstra
85ae77c108
tilelink2 RAMModule: carefully stage the pipeline to make BRAMs possible
2016-09-12 10:32:25 -07:00
Wesley W. Terpstra
9560df537c
tilelink2 RegisterRouter: allow sub-4k devices in order to make useful unit tests
2016-09-12 10:32:24 -07:00
Wesley W. Terpstra
26f9e2dfbd
tilelink2 Parameters: fix width=1 address truncation bug
2016-09-12 10:32:24 -07:00
Wesley W. Terpstra
98a4facac7
tilelink2 RAMModel: clear Mems on power-up
2016-09-12 10:32:24 -07:00
Wesley W. Terpstra
17f7ab18de
tilelink2 RAMModel: model the state a RAM would have for Put+Gets
2016-09-12 10:32:24 -07:00
Wesley W. Terpstra
488b93d146
tilelink2 Parameters: if you support PutPartial, you must support PutFull
2016-09-12 10:32:24 -07:00
Wesley W. Terpstra
d6261e8ce8
tilelink2 Edge: add a numBeats1 method for predecremented code
2016-09-12 10:32:24 -07:00
Wesley W. Terpstra
5604049927
tilelink2 Buffer: support an unlimited number of channels
2016-09-12 10:32:24 -07:00
Yunsup Lee
d985cdfc66
Merge branch 'master' into refactor-periphery
2016-09-10 23:42:13 -07:00
Yunsup Lee
fea31c7061
let GlobalAddrMap and ConfigString overridable
2016-09-10 23:39:44 -07:00
Yunsup Lee
bb3f514e8d
now able to add periphery devices through traits
...
Unfortunately, I had to touch a lot of code, which weren't quite possible to split up into multiple commits.
This commit gets rid of the "extra" infrastructure to add periphery devices into Top.
2016-09-10 23:39:29 -07:00
Megan Wachs
77e4aa63f8
Get rid of the unecessary Parameters for Async Reset Reg
2016-09-09 16:24:35 -07:00
Andrew Waterman
b695ab5292
Merge branch 'master' into tweaks
2016-09-09 15:04:21 -07:00
Megan Wachs
5f5989848c
Merge remote-tracking branch 'origin/master' into black_box_regs
2016-09-09 13:12:52 -07:00
Andrew Waterman
656aa78f7d
Pipeline FMAs more deeply by default
...
Rocket's QoR has improved enough that the FMAs are on the critical
path. This change seems to keep the integer pipeline's logic
paths balanced with the FPU.
2016-09-09 11:06:42 -07:00
Andrew Waterman
eaa4b04ee5
Check D$ store->load collisions more precisely
...
Tolerate, for example, a half-word store and a half-word load to
different halves of the same word.
2016-09-09 11:06:42 -07:00
Wesley W. Terpstra
c28ca37944
tilelink2: get rid of fragile implicit lazyModule pattern, and support :=
...
We can more reliably find the current LazyModule from the LazyModule.stack
2016-09-08 23:06:59 -07:00
Wesley W. Terpstra
b587a409a3
tilelink2 Node: make it possible for {Identity,Output,Input}Node to pass a Vec
...
In order to implement a pass-through RAM Monitor model, we will want to support
a variable number of inputs and outputs with BOTH different manager and client
parameters on each bundle.
2016-09-08 21:34:20 -07:00
Wesley W. Terpstra
48ca478578
Merge branch 'master' into intbar
2016-09-08 21:09:59 -07:00
Wesley W. Terpstra
808a7f60f4
tilelink2 Legacy: it's only an error if it's valid ( #264 )
2016-09-08 21:09:40 -07:00
Megan Wachs
fda4c2bd76
Add a way to create Async Reset Registers and a way to easily access them with TL2
2016-09-08 20:02:07 -07:00
Megan Wachs
c1eb1f12a2
tilelink2: Rename GPIO to Example to avoid conflicts with real GPIO devices
2016-09-08 20:02:07 -07:00
Wesley W. Terpstra
cbf0670156
tilelink2 Legacy: it's only an error if it's valid
2016-09-08 19:32:00 -07:00
Wesley W. Terpstra
1b07d53f70
tilelink2 IntNodes: record interrupt ranges in parameters
2016-09-08 18:51:43 -07:00
Wesley W. Terpstra
66f58cf2d0
tilelink2 RegisterRouter: support new TL2 interrupts
2016-09-08 15:25:50 -07:00
Wesley W. Terpstra
23e896ed5d
tilelink2 IntNodes: support interrupt graphs
2016-09-08 15:25:48 -07:00
Wesley W. Terpstra
d7df7d3109
tilelink2: connect Nodes to LazyModules for better error messages
2016-09-08 15:24:04 -07:00
Wesley W. Terpstra
53987cd9d4
tilelink2 Nodes: support non-Bundle data for io type
2016-09-08 15:19:12 -07:00
Wesley W. Terpstra
60a503dc2f
tilelink2 RegField: add a w1ToClear RegField
2016-09-08 14:02:49 -07:00
Wesley W. Terpstra
99b7e734cd
tilelink2 Bundles: fix wrong sink width!
2016-09-08 13:47:40 -07:00
Wesley W. Terpstra
9bfd8c1cf5
TL2 WidthWidget ( #258 )
...
* tilelink2 Narrower: support widenening and narrowing on all channels
Be extra careful with the mask transformations
We need to make sure that narrowing or widening do not cause a loss
of information about the operation. The addr_hi+(mask|addr_lo) conversions
are now 1-1, except on D, which should not matter.
* tilelink2 SRAM: work around firrtl SeqMem bug
* tilelink2 WidthWidget: renamed from Narrower (it now converts both ways)
* tilelink2 mask: fix an issue with width=1 data buses
2016-09-08 10:38:38 -07:00
Yunsup Lee
2c000a99da
compartmentalize Top into periphery traits
2016-09-08 02:08:57 -07:00
Yunsup Lee
e35e7b2ee3
Fix routing in non-contiguous MMIO regions
...
This is a temporary fix, which can generate more hardware than necessary, but this is OK for now, since this code will soon be replaced with tilelink2 code.
2016-09-07 19:28:12 -07:00
Andrew Waterman
7603b86239
Merge branch 'master' into use-companion
2016-09-07 12:56:55 -07:00
Colin Schmidt
254f49093c
only use companion objects for types
2016-09-07 12:32:34 -07:00
Andrew Waterman
23d0b31615
Merge branch 'master' into tilelink2.2
2016-09-07 11:47:50 -07:00
Andrew Waterman
02a2439222
Support a degenerate PLIC with no interrupts
...
Resolves #249
2016-09-07 11:21:13 -07:00
Andrew Waterman
70cfd7ce13
Make DefaultRV32Config be RV32IMAFCS, not RV32IMC
...
The latter is more the domain of TinyConfig.
2016-09-07 01:58:25 -07:00
Andrew Waterman
a7f47f3c23
Reduce default BTB size
...
The old value 62 seems to have been a typo introduced over 2 years ago
in commit 63bd0b9d2a
. The intent was to
fit the dhrystone working set (rofl) which the new value of 40 does.
2016-09-07 01:51:27 -07:00
Andrew Waterman
9fea4c83da
Add RV32F support
2016-09-07 00:05:39 -07:00
Andrew Waterman
66e9f027e0
Add MuxT to mux on Tuple2 and Tuple3
2016-09-07 00:05:38 -07:00
Andrew Waterman
511cc6c5c5
Evaluate arg to Boolean.option lazily
2016-09-07 00:05:38 -07:00
Andrew Waterman
a0dcd42e80
avoid erroneously setting tags valid during flush
2016-09-07 00:05:38 -07:00
Yunsup Lee
fb05f5a07f
remove parameter ExtIOAddrMapEntries ( #250 )
...
with the AddrMap ordering constraint relaxed, this parameter is no longer needed.
2016-09-07 00:05:00 -07:00
Wesley W. Terpstra
d2421654c4
tilelink2: refactor address into addr_hi on ABC and addr_lo on CD
...
We need addr_lo in order to properly convert widths.
As part of the refactoring, move all methods out of the Bundles
2016-09-06 23:46:44 -07:00
Yunsup Lee
b76612f357
relax contraint on adding AddrMapEntry to AddrMap ( #248 )
...
now you can add them in any order. there's an explicit check at the end to figure out whether there are overlapping regions.
2016-09-06 21:53:55 -07:00
Wesley W. Terpstra
aae4230627
tilelink2: fix bugs found by Megan in Legacy converter
2016-09-06 13:12:33 -07:00
Yunsup Lee
56d81b0034
fix configstring printout with no memory
2016-09-06 10:40:11 -07:00
Wesley W. Terpstra
54ab14cd9d
tilelink2: statically optimize numBeats for simple managers
2016-09-05 22:11:03 -07:00
Wesley W. Terpstra
314d6ebd6f
tilelink2: stricter TransferSizes requirements
2016-09-05 22:10:28 -07:00
Wesley W. Terpstra
56170c605c
tilelink2: be more forgiving in what Legacy TL requires
2016-09-05 21:12:51 -07:00
Wesley W. Terpstra
3167539331
tilelink2: Narrower must be little-endian
2016-09-05 20:58:41 -07:00
Wesley W. Terpstra
ded246fb95
tilelink2: relax max transfer size; the real requirement is not exceeding alignment
2016-09-05 20:58:41 -07:00
Wesley W. Terpstra
cf0291061d
tilelink2: fix a bug in UIntToOH1 triggered if the size was too big
2016-09-05 20:58:41 -07:00
Wesley W. Terpstra
9f45212c95
tilelink2: Fragmenter needs to update subaddress
2016-09-05 20:58:41 -07:00
Wesley W. Terpstra
757d46279e
tilelink2: expand data correctly in D channel narrower
2016-09-05 20:58:41 -07:00
Wesley W. Terpstra
0faa8c4051
tilelink2: fix Xbar bug where Mux1H broke FSM if only one manager
2016-09-05 20:58:41 -07:00
Wesley W. Terpstra
a0c25880c7
tilelink2: Monitor should check mask of reconstructed request
2016-09-05 20:58:41 -07:00