1
0
Commit Graph

756 Commits

Author SHA1 Message Date
mwachs5
f9e0a7ac24 Merge branch 'master' into async_register_crossing 2016-09-27 15:54:34 -07:00
Wesley W. Terpstra
3926cb936b rocketchip: add pbus width and AMO With classes (#357) 2016-09-27 15:52:13 -07:00
Wesley W. Terpstra
eaea138d0d tilelink2: don't use chisel3 namespace (#355) 2016-09-27 14:44:26 -07:00
Henry Cook
f5502df6ab Merge branch 'master' into async_register_crossing 2016-09-27 14:08:27 -07:00
Wesley W. Terpstra
357d06ac9c tilelink2 WidthWidget: Gets must have their mask adjusted (#353)
The mask of a Get should also be converted.

This manifested as a bug when going from 32=>64 bits. A large Get
could end up with mask that was not full.
2016-09-27 14:06:02 -07:00
Megan Wachs
3ce08f40a5 crossing: Remove reset from the logic in Register Crossing because it is no longer needed when the underlying crossings are asynchronously reset. Update the order of operations 2016-09-27 13:36:28 -07:00
Howard Mao
71a9c78e4b add WidthAdapter from AXI slave to Coreplex TL slave 2016-09-27 12:48:01 -07:00
Howard Mao
7d6fb950b6 Give TileLink IDs more sensible names
* Outermost -> MCtoEdge
 * MMIO_Outermost -> MMIOtoEdge

Then the corresponding parameters objects are

 * L1toL2 -> innerParams
 * L2toMC -> outerMemParams
 * L2toMMIO -> outerMMIOParams
 * MCtoEdge -> edgeMemParams
 * MMIOtoEdge -> edgeMMIOParams
2016-09-27 12:48:01 -07:00
Howard Mao
8a55521b01 move memory width adapter from coreplex to periphery 2016-09-27 12:48:01 -07:00
Howard Mao
e36441a046 use correct parameters object for MMIO width adapter 2016-09-27 12:48:01 -07:00
Howard Mao
201e247f73 Factor coreplex IO connection into separate trait (#350)
This would allow, for instance, putting the coreplex on a separate clock
domain and crossing the IOs over through asynchronous queues.

The ExampleMultiClockTop* classes are removed since they no longer fit
into the class hierarchy.
2016-09-27 11:55:32 -07:00
Howard Mao
6316ebd58f make naming of L2toMC parameter object consistent between coreplex and periphery 2016-09-26 17:28:21 -07:00
Howard Mao
ea9f0a868f TileLink utility objects should not take implicit parameters
We have a handful of TileLink-related helper objects
(wrappers, unwrappers, width adapters, and enqueuers). Previously, using
them could be error-prone, because you had to make sure the implicit
parameters they took in had the same TLId as the TileLinkIO bundles
passed in as inputs. This is rather silly, we should just use the
parameters in the bundle.
2016-09-26 17:28:21 -07:00
Howard Mao
803739a95c Make sure coreplex mmio's TLId is correct (thanks to zizztux) 2016-09-26 17:28:21 -07:00
Howard Mao
c741ada619 get TraceGen working again 2016-09-26 17:28:21 -07:00
Wesley W. Terpstra
d9e209365d Tl2 addr width0 (#346)
* tilelink2 Edges: add accessor methods for address and addr_{hi,lo}

* tilelink2: use addr_lo instead of relying on truncation

Truncation can mess up if the width should be 0, but IS 1.
2016-09-26 17:00:03 -07:00
Wesley W. Terpstra
72c205b54f tilelink2 AddressSet: add .misaligned(low, size) helper method (#345)
This helps devices with misaligned ranges still connect to TL2.
2016-09-26 16:01:09 -07:00
Wesley W. Terpstra
dd9558f45d rocketchip: generate GraphML output 2016-09-26 14:35:46 -07:00
Wesley W. Terpstra
1773eb4405 tilelink2 LazyModule: output GraphML of the bus 2016-09-26 14:35:46 -07:00
Wesley W. Terpstra
35da9320bc tilelink2 Nodes: expose connectivity in RootNode 2016-09-26 14:35:46 -07:00
Wesley W. Terpstra
14cd39e045 rocketchip: rename identically names devices with _%d (#340)
* rocketchip: rename identically names devices with _%d

If you connect two devices with the same name in TL2 (totally ok there),
when they get put into the TL1 addrmap, one gets silently overwritten.
This renames the second occurance as _1, third as _2, and so on...

* junctions: blow if duplicates add to addrmap
2016-09-26 13:05:49 -07:00
mwachs5
77a0f76289 Cleanup jtag dtm (#342)
* debug: Clean up Debug TransportModule synchronizer

With async reset async queues, I feel its safe/cleaner
to remove the one-off "AsyncMailbox verilog black-box
and use the common primitive.

I also added some comments about correct usage of this
block. Probably the 'TRST' signal should be renamed
to make it less confusing, as it requires some processing
of the real JTAG 'TRST' signal.
2016-09-26 11:10:27 -07:00
mwachs5
8641639873 Async rst async queue (#336)
* crossing: use async reset

* crossings: asyncqueue needs Asynchronous reset.

* crossing: Actually enable the head of the synchronizer flop chain

* crossing: remove reset from logic. This flop will no longer be written during reset because valid will be low.

* crossing: Tidy up code  & comments
2016-09-26 11:08:38 -07:00
Wesley W. Terpstra
d787bae0d0 tilelink2 Xbar: decouple ready from valid (#338)
This moves the Xbar from using custom code to using the Arbiter.
The arbiter has better ready-valid decoupling.
2016-09-23 16:24:29 -07:00
Wesley W. Terpstra
d175bb314d Periphery: make bus width and arithmetic atomics configurable (#337) 2016-09-23 15:25:58 -07:00
Wesley W. Terpstra
47843d8ec1 tilelink2: maxLgSize should be accurate (#332) 2016-09-22 22:06:22 -07:00
Wesley W. Terpstra
c5706afc11 RegField: remove obsolete split method
This is now natively supported by the regmap(...) invocation.
2016-09-22 20:52:47 -07:00
Wesley W. Terpstra
fc44151f10 RegField: add name and description fields
In the future we can generate interesting documentation and headers.
2016-09-22 20:52:46 -07:00
Wesley W. Terpstra
5e34b313ee RegMapper: regmap(...) now takes BYTE addresses
If a device has configurable bus-width, we need a stable way of
enumerating registers. The byte offset stays unchanged.

This change also makes it possible to put an arbitrary number of RegFields
starting at some address which are then chopped up into appropriately bus-
sized registers.
2016-09-22 20:52:46 -07:00
Wesley W. Terpstra
972ca06729 RegField: remove RegField.bytes; it was dangerous
The implementation unconditionally drove the register.
This made it incompatible with drivers from the device itself.
Besides, writing only parts of a register at a time is ultra-shady.
2016-09-22 20:52:46 -07:00
Wesley W. Terpstra
a421469754 tilelink2: change adapters to use TLAdapter(params, defaults)(node)
This API makes it much more readable when you have multiple adapters
combined into a single line. The arguments for each adapter stay
beside the adapter.

For example, this:
  peripheryBus.node := TLWidthWidget(TLBuffer(TLAtomicAutomata()(TLHintHandler(legacy.node))), legacy.tlDataBytes)
becomes this:
  peripheryBus.node := TLWidthWidget(legacy.tlDataBytes)(TLBuffer()(TLAtomicAutomata()(TLHintHandler()(legacy.node))))
2016-09-22 20:52:46 -07:00
Howard Mao
22053289ef fix typo rv64iu -> rv64ui 2016-09-22 17:33:35 -07:00
Henry Cook
673efb400d Merge branch 'master' into unittest-config 2016-09-22 16:20:53 -07:00
Henry Cook
1e54820f8c Merge remote-tracking branch 'origin/master' into unittest-config 2016-09-22 16:03:51 -07:00
Henry Cook
411ee378de Provide a GeneratorApp object per user package. Extract RocketTestSuite from coreplex into rocketchip and provide GeneratorApp defaults for other target packages. 2016-09-22 15:59:29 -07:00
Henry Cook
391be8d740 tilelink2 RegisterRouter: minLatency is never more than 1 2016-09-22 15:51:15 -07:00
Wesley W. Terpstra
a3e88fa13a tilelink2 Atomics: optimize the sign-extension circuit 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
9f1f6fc61f Comparator: tolerate mismatched data when it is undefined 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
ed038678ef tilelink2 Fuzzer: work around for firrtl/verilator performance issue
Big Vec()s cause very slow compilation.
2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
1e7480b6fc tilelink2 Monitor: work around for firrtl/verilator performance issue
Big Vec()s cause problems for these tools.
2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
ec2030df31 tilelink2 Legacy: convert TL1 atomic operand size 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
0a3718881f rocketchip: re-enable testing of atomics 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
e5da3eb8bb tilelink2 Atomics: support arithmetic atomics 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
5b80fe5b51 tilelink2 Atomics: support Logical AMOs 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
4066fbe18f tilelink2 RAMModel: exploit latency to remove bypass 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
e0ade8c5a9 tilelink2 Atomics: exploit minLatency to eliminate bypass 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
3bb2580223 tilelink2 Monitor: detect minLatency violations 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
2b24c4b1b4 tilelink2: most adapters can wipe away latency 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
c115913624 tilelink2 Buffer: increase the minLatency on ports 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
05beb20dc4 tilelink2: specify the minLatency for SRAM+RR 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
44277c1db3 tilelink2 Parameters: include a minLatency parameter for optimization 2016-09-22 15:18:54 -07:00
Wesley W. Terpstra
cf39c32b0e tilelink2 Fuzzer: test Atomics 2016-09-22 15:18:53 -07:00
Wesley W. Terpstra
2b9403633d tilelink2 RAMModel: support (by ignoring) atomics 2016-09-22 15:18:53 -07:00
Wesley W. Terpstra
ce204f604a tilelink2 AtomicAutomata: prototype flow control complete 2016-09-22 15:18:53 -07:00
Wesley W. Terpstra
42b10356fa tilelink2: add a general-purpose Arbiter 2016-09-22 15:18:53 -07:00
Wesley W. Terpstra
7636e772c8 tilelink2 Fuzzer: only generate legal atomics 2016-09-22 15:18:53 -07:00
Wesley W. Terpstra
f5d604d8f8 tilelink2 Parameters: poison ports with unsafe atomics
We need to detect if an AtomicAutomata's output ever gets mixed
with some other source of operations.
2016-09-22 15:18:53 -07:00
Wesley W. Terpstra
d1151e2f0f tilelink2 Nodes: split connect into eager and lazy halves 2016-09-22 15:18:50 -07:00
Wesley W. Terpstra
684072023f tilelink2 Monitor: make it a LazyModule in the hierarchy 2016-09-22 15:14:20 -07:00
Wesley W. Terpstra
def497861b tilelink2 Bundles: add 1-way snoop bundles 2016-09-22 15:14:20 -07:00
Wesley W. Terpstra
69a1f8cd1f tilelink2 Monitor: detect if sources are mishandled 2016-09-22 15:14:19 -07:00
Henry Cook
83c08a931d [WIP] Generators for unittest and groundtest; disambiguate groundtest.TrafficGenerator 2016-09-22 14:57:18 -07:00
Henry Cook
47c5d1a992 [WIP] Move RocketTestSuite generation into RocketchipGenerator 2016-09-22 14:31:45 -07:00
Albert Ou
d76b762657 tilelink2 Fragmenter: Mask low bits of D channel addr_lo
This fixes an issue where passing addr_lo through unchanged triggered
unaligned address assertions in the Monitor.
2016-09-22 12:36:28 -07:00
Howard Mao
cd96a66ba6 replace verilog clock divider with one written in Chisel 2016-09-22 11:32:29 -07:00
Howard Mao
cbd702e48e make sure junctions and uncore unittests both run 2016-09-21 20:17:52 -07:00
Yunsup Lee
1b1ef3be07 simplify base Coreplex bundle 2016-09-21 18:29:28 -07:00
Yunsup Lee
d2df6397cd rename trc (tile reset clock) bundles to tcr (tile clock reset) 2016-09-21 18:29:28 -07:00
Yunsup Lee
5bb575ef74 rename internal/external MMIO network to cbus/pbus respectively 2016-09-21 18:29:28 -07:00
Henry Cook
64fe010369 [unittest] Config import tweaks 2016-09-21 17:40:39 -07:00
Henry Cook
fd5e00fed9 [coreplex] rename Testing.scala -> RocketTestSuite.scala 2016-09-21 17:35:39 -07:00
Henry Cook
270011b768 [unittest] more Config cleanup 2016-09-21 17:30:14 -07:00
Yunsup Lee
7afd630d3e add multiclock support to Coreplex 2016-09-21 16:55:26 -07:00
Andrew Waterman
8e63f4a1a5 Remove ClockToSignal and vice-versa
Clock.asUInt and Bool.asClock now suffice.
2016-09-21 16:17:14 -07:00
Henry Cook
335e866176 [unittest] Parallelize UnitTestSuite (#319)
* [unittest] Parallelize UnitTestSuite so all tests have their own timer, runs until all finish or any timeout. Adds SimpleTimer.

* [util] Timer spacing cleanup

* [unittest] Remove Config reference to UnitTestTimeout
2016-09-21 13:05:22 -07:00
Andrew Waterman
12d0c00822 Fix mtime RegField handling
RegField.bytes was unconditionally overwriting mtime, preventing it
from ever ticking.  Avoid RegField.bytes by splitting mtime into
a Seq of words.
2016-09-20 15:00:52 -07:00
Henry Cook
40f6f31611 [unittest] further refactor unittest framework 2016-09-20 14:14:30 -07:00
Henry Cook
ed91e9a89b Merge remote-tracking branch 'origin' into testharness-refactor 2016-09-20 13:03:21 -07:00
Henry Cook
b97a0947a9 [rocketchip] enable piecewise Generator output 2016-09-20 12:57:56 -07:00
Yunsup Lee
1a09e46f69 Merge branch 'master' into fix-addrmap-error-msg 2016-09-19 18:08:58 -07:00
Andrew Waterman
3b38736a8e Make BaseTopModule and BaseTopModule abstract
They aren't meant to be directly instantiated.
2016-09-19 17:18:35 -07:00
Andrew Waterman
d0572d6aab Allow reset vector to be set dynamically
A chip's power-up sequence, or awake-from-sleep sequence, may wish to
set the reset PC based upon dynamic properties, e.g., the settings of
external pins.  Support this by passing the reset vector to the Coreplex.
ExampleTop simply hard-wires the reset vector, as was the case before.

Additionally, allow MTVEC to *not* be reset.  In most cases, including
riscv-tests, pk, and bbl, overriding MTVEC is one of the first things
that the boot sequence does.  So the reset value is superfluous.
2016-09-19 17:18:03 -07:00
Andrew Waterman
e6c1bcfedd Expose carry-out bits from WideCounter 2016-09-19 15:54:17 -07:00
Yunsup Lee
1b26d78114 correctly print out the addrmap overlapping error message 2016-09-19 13:34:58 -07:00
Henry Cook
df442ed82c [rocketchip] avoid pending merge conflict] 2016-09-19 13:24:01 -07:00
Henry Cook
ddcf1b4099 Use PROJECT rather than MODEL in name of binary and generated src files. 2016-09-19 13:23:17 -07:00
Henry Cook
7b8aa6c839 [rocketchip] split out Base and Example tops 2016-09-19 11:00:13 -07:00
Andrew Waterman
a49814c667 Allow WideCounter to not be reset 2016-09-18 18:45:51 -07:00
Wesley W. Terpstra
9817a00ed9 tilelink2: Fuzzer should check address validity before injection 2016-09-17 17:07:21 -07:00
Wesley W. Terpstra
b11839f5a1 tilelink2: differentiate fast/safe address lookup cases 2016-09-17 17:04:18 -07:00
Wesley W. Terpstra
b4baae4214 tilelink2: minimize Xbar decode logic 2016-09-17 16:14:25 -07:00
Wesley W. Terpstra
76d8ed6a69 tilelink2: remove 'strided'; !contiguous is clearer 2016-09-17 16:14:25 -07:00
Wesley W. Terpstra
fa0f119f3c tilelink2: consider the implications of negative address mask 2016-09-17 16:14:22 -07:00
Wesley W. Terpstra
e437508548 tilelink2: track interrupt connectivity like in TL2 2016-09-17 14:43:48 -07:00
Wesley W. Terpstra
01c1886b9d Utils: cacheable only if there is a cache manager 2016-09-17 00:56:21 -07:00
Wesley W. Terpstra
6c3269a1d8 SRAM: optionally (default: true) executable 2016-09-17 00:19:37 -07:00
Wesley W. Terpstra
e749558190 ROM: optionally (default: true) executable 2016-09-17 00:19:09 -07:00
Wesley W. Terpstra
c70045b8b3 Utils: express cacheability from TL2 to TL1 2016-09-17 00:16:40 -07:00
Wesley W. Terpstra
e3d2bd3323 Top: print memory region properties, RWX [C] 2016-09-17 00:16:00 -07:00
Wesley W. Terpstra
5c858685aa Utils: support managers with multiple addresses 2016-09-16 18:03:49 -07:00