- Feb 25, 2020
-
-
Bradley Morrell authored
-
Bradley Morrell authored
-
- Feb 21, 2020
-
-
Bradley Morrell authored
-
Bradley Morrell authored
-
- Feb 20, 2020
-
-
Bradley Morrell authored
-
Bradley Morrell authored
-
- Feb 19, 2020
-
-
Bradley Morrell authored
-
- Feb 11, 2020
-
-
Bradley Morrell authored
-
- Feb 04, 2020
-
-
Bradley Morrell authored
-
- Jan 31, 2020
-
-
Bradley Morrell authored
-
Bradley Morrell authored
-
- Jan 28, 2020
-
-
Dhruv Gupta authored
-
Deborah Soung authored
-
- Jan 25, 2020
-
-
Adam Izraelevitz authored
* Bump chisel3 3.2.x for Select.instances bugfix Co-authored-by:Deborah Soung <debs@sifive.com>
-
- Jan 23, 2020
-
-
Henry Cook authored
* clocks: copy clocks package from sifive.fpgashells; update imports * clocks: bundles can use chisel3 bundle to get cloneType for free * clocks: use chisel3 in all files * clocks: add ClockSinkDomain and ClockSourceDomain * clocks: allow for unspecified clock freqs * clocks: add TestClockSource test harness clock generators * clocks: domain nodes are named clockNode * clocks: source and sink nodes offer fixedClockResources * rename package from clocks to prci * clocks: ClockGroup.pickGroupFromCrossingType * prci: make abstract base ClockDomain class * prci: add FixedClockBroadcastNode * tilelink: add fixedClockOpt to HasTLBusParams * tilelink: add clock nodes to TLBusWrapper * subsystem: add clockGroupNode to BaseSubsystem * devices: add example device that can gate clock from TL register-mapped interface * prci: add draft of ClockDivider adapter * buswrapper: derive dtsClk from clockNode.fixedClockResources * util: update ClockDivider to chisel3 * prci: add package object to hold asyncMux helper func * prci: add ClockGroupSourceNode and ClockGroupSinkNode * prci: ClockGroup and ClockGroupBroadcast are LazyRawModuleImps * prci: domain creates clock and reset output IOs * prci: rename ClockGroupBroadcastNode to ClockGroupAggregateNode * tilelink: TLBusWrapper includes ClockGroupAggregator * tilelink: TLBusWrapper crossTo/FromBus take implicit ClockGroupIdentityNode * system: ExampleRocketSystem supplies default simple clock source * subsystem: BaseSubsystem has ClockGroupIdentityNode * prci: SimpleClockGroupSource can have a variable number of sources * groundtest: subsystem needs dummyClockGroupSourceNode * prci: add FixedClockBroadcast helper module and use in TLBusWrapper * prci: ClockEphemeralNode and ClockGroupEphemeralNode * prci: TLBusWrapper uses ClockGroupEphemeralNode * subsystem: add traits HasAsyncClockGroupsNode and Attachable * subsystem: clean up Attachable and add SubsystemDriveAsyncClockGroupsKey * prci: add PRCIClockGroupNode: NexusNode(ClockGroupImp) * tilelink: better name for BuwWrapper clock aggration nodes * groundtest: delete import that was actually unused * prci: move ClockGroupDriverParameters to prci
-
- Jan 22, 2020
-
-
Ernie Edgar authored
Debug SBA: Don't reset sbState if dmactive is set to 0
-
- Jan 21, 2020
-
-
Ernie Edgar authored
-
Dhruv Gupta authored
-
- Jan 15, 2020
-
-
Richard Xia authored
Correct spelling of TileLink in Object Model protocol description
-
Richard Xia authored
-
- Jan 14, 2020
-
-
John Ingalls authored
Rocket Core DCache TileLink: require FIFO to allVolatile regions
-
John Ingalls authored
-
John Ingalls authored
GroundTestTileParams Some(ICacheParams)
-
- Jan 13, 2020
-
-
John Ingalls authored
-
- Jan 12, 2020
-
-
Ernie Edgar authored
-
- Jan 11, 2020
-
-
Srivatsa Yogendra authored
-
John Ingalls authored
Rocket Core DCache TileLink: dont require Fifo just yet
-
Ernie Edgar authored
-
Richard Xia authored
Fix offsets of DMOuter registers in Object Model.
-
John Ingalls authored
-
John Ingalls authored
TLFIFOFixer.Policy requireFifo
-
- Jan 10, 2020
-
-
Dhruv Gupta authored
-
Dhruv Gupta authored
-
Richard Xia authored
The previous Object Model implementation did not account for the relative offset of the dmOuter register address block. This commit adds some logic to translate the offsets into absolute addresses.
-
John Ingalls authored
-
John Ingalls authored
-
- Jan 09, 2020
-
-
Andrew Waterman authored
-
- Jan 08, 2020
-
-
Dhruv Gupta authored
-
Dhruv Gupta authored
-
- Jan 07, 2020
-
-
Andrew Waterman authored
Like 37a3161b, this is also just for post-silicon debug purposes. Disabling prefetch doesn't add much extra PSD benefit beyond disabling speculative refill, because the latter disables prefetching on speculative accesses already, leaving only next-line prefetches following non-speculative refills. But it is worth something.
-