summaryrefslogtreecommitdiff
path: root/src/main
AgeCommit message (Collapse)Author
2015-08-04Clean up MuxAndrew Waterman
2015-08-04Fix inferred width of ReverseAndrew Waterman
2015-08-04Fix Fill on BoolAndrew Waterman
2015-08-04Reg(init=UInt(0,N)) should force Reg width to NAndrew Waterman
I don't really like the syntax, but it's for backwards compatibility.
2015-08-04Refactor Bundle field enumerationAndrew Waterman
2015-08-04Add better cloneTypeWidth and deprecate old oneAndrew Waterman
2015-08-04Don't create zero-width Enums, at least for nowAndrew Waterman
2015-08-03improve macro hygieneHenry Cook
2015-08-03Improve type-safety of switch/is blocks using macros and SwitchContext. ↵Henry Cook
Removed switch key stack.
2015-08-03Fix << width bugAndrew Waterman
2015-08-03Don't zero-initialize for Wire(init=...)Andrew Waterman
2015-08-03No more shitballsAndrew Waterman
2015-08-02shitballsAndrew Waterman
2015-08-02Move comparison operators to UInt/SIntAndrew Waterman
2015-08-02Remove legacy Params stuffAndrew Waterman
If we need anything from these files, we can re-add it later.
2015-08-02soften permissions on dirVarHenry Cook
2015-08-02Work around FIRRTL initialization pedantryAndrew Waterman
Initialize all wires, output ports, and instance input ports to 0.
2015-08-02Work around FIRRTL accessor problemAndrew Waterman
2015-08-02Use flatMap instead of map/reduceAndrew Waterman
2015-08-01Compute node directions correctlyAndrew Waterman
2015-08-01When reflecting, sort fields for determinismAndrew Waterman
The damn JVM isn't deterministic.
2015-08-01Clean up flip/asInput/asOutputAndrew Waterman
2015-07-31Disallow dynamic bit range extractionAndrew Waterman
It wasn't correctly implemented, anyway.
2015-07-31Improve handling of := and <>Andrew Waterman
- Check that ground types match - Emit FIRRTL bulk-connect instead of := for bundle connections
2015-07-31Implement getWidth more completely and less buggilyAndrew Waterman
2015-07-31Open heart surgery on IDs/namingAndrew Waterman
This commit institutes a more structured system for managing namespaces. It fixes a handful of namespace collision bugs, and also improves performance by abandoning ungainly string-based IDs.
2015-07-30Work around FIRRTL literal restrictionsAndrew Waterman
2015-07-30Better literal checkingAndrew Waterman
2015-07-30Check for negative UInt literalsAndrew Waterman
2015-07-30Correct implementation of andRAndrew Waterman
2015-07-30Make Vec.fill(n)(x) the same as Vec(x, n)Andrew Waterman
2015-07-30Emit clocks and resetsAndrew Waterman
2015-07-30Add missing Wire()Andrew Waterman
2015-07-30Move towards compatibility with FIRRTL 0.1.3Andrew Waterman
2015-07-29Remove most operators from BitsAndrew Waterman
The concrete return type was the type of the LHS, which is wrong for most mixed-signedness code. We can get it right using double-dispatch, but let's wait until we know we need to do so. I left the comparison operators because their return type is always Bool.
2015-07-29For Mux1H, use UInt instead of BitsAndrew Waterman
2015-07-29Remove nondeterminism in field namingAndrew Waterman
We must compensate for Class.getMethod's nondeterministic ordering.
2015-07-29Add SInt-by-UInt multiplication operatorAndrew Waterman
2015-07-29Add newline at end of .fir fileAndrew Waterman
2015-07-29Print out basic status information when elaboratingAndrew Waterman
2015-07-29Fix Bundle port orderingAndrew Waterman
2015-07-29Use Seq, not Iterable, when traversal order mattersAndrew Waterman
2015-07-29Clean up Reg constructorAndrew Waterman
2015-07-28toBits and fromBits must be inverse operationsAndrew Waterman
For Vec and Bundle, toBits and fromBits got the flattening order wrong and so weren't inverses of each other.
2015-07-28Avoid some heap allocationsAndrew Waterman
2015-07-28Speed up code emission using StringBuildersAndrew Waterman
2015-07-28Rename sumLog2Width to sumPow2WidthAndrew Waterman
2015-07-28Avoid needless Vec instantiationAndrew Waterman
This avoids some cases where cloneType would need to be defined.
2015-07-28Improve cloneType error messagesAndrew Waterman
2015-07-28Only attempt to name vals, not defsAndrew Waterman