| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
Resolves #917
|
|
Quickfix for LoadMemoryTransform that gets this to work with Instance
Annotations. The new Instance Annotations caused a corner case where a
LoadMemoryAnnotation would be duplicated (via update/renaming) and the
resulting annotation would differ from the original in only their
originalMemoryNameOpt field. This corrects that by having the
ChiselLoadMemoryAnnotation also emit the originalMemoryNameOpt field where
it did not previously.
First part of a fix for freechipsproject/firrtl#922.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
* Added new strongly-typed enum construct called "StrongEnum". "StrongEnum" will automatically generate annotations that HDL backends can use to mark components as enums
Removed "override val width" constructor parameter from "Element" so that classes with variable widths, like the new strong enums, can inherit from it
Changed the parameter types of certain functions, such as "switch", "is", and "LitArg.bindLitArg" from "Bits" to "Element", so that they can take the new strong enums as arguments
* Added tests for the new strong enums
* Changed StrongEnum exception names and made sure in StrongEnum tests that the correct types of exceptions are thrown
* Fixed bug where an enum's global annotation would not be set if it was used in multiple circuits
Made styling changes to StrongEnum.scala
* Reverted accidental changes to the AnnotatingDiamond test
* Changed the API for casting non-literal UInts to enums
Added an isValid function that checks whether or not enums have valid values
Calling getWidth on an enum's companion object now returns a BigInt instead of an Int
* Casting a literal to an enum using the StrongEnum.castFromNonLit(n) function is now simply a wrapper for StrongEnum.apply(n)
* Fixed compilation bug
* * Added "next" method to EnumType
* Renamed "castFromNonLit" to "fromBits"
* The FSM example in the test/scala/cookbook now uses StrongEnums
* * Changed strong enum API, so that users no longer have to declare both a class and a companion object for each strong enum
* Strong enums do not have to be static any longer
* * Added scope protections to ChiselEnum.Value so that users cannot call it
outside of a ChiselEnum definition
* Renamed ChiselEnum.Value type to ChiselEnum.Type so that we can give
it a companion object just like UInt and Bool do
* * Moved strong enums into experimental package
* Non-literal UInts can now be cast to enums with apply() rather than
fromBits()
* Reduced code-duplication by moving some functions from EnumType and
Bits to Element
|
|
* Modify noenq to set the input wires to DontCare.
|
|
This places all do_* methods (and two unary methods in SInt and FixedPoint
that act like do_* methods) inside the ScalaDoc group
"SourceInfoTransformMacro". Classes/objects which need information about
this group have an additional bare trait mixed in, `SourceInfoDoc`, that
provides information about the group and its priority.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
* Ability to load memories at simulation startup
* first pass
* create annotation
* create skeleton Transform
* Work in progress
Building out transform and pass now
* Support for LoadMemory annotation
* Creates chisel and firrtl LoadMemory annotations
* LoadMemoryTransform converts annotation into BlackBox InLine
* Simple test that verilog bound modules get created.
* Support for LoadMemory annotation
* Supports Bundled/multi-field memories
* more tests
* support for `$readmemh` and `$readmemb`
* warns if suffix used in file specification.
* Support for LoadMemory annotation
* Use standard chisel annotation idiom
* Support for LoadMemory annotation
* Fixes for @seldridge nits and super-nits
* Support for LoadMemory annotation
- transform now only runs if emitter is an instance of VerilogEmitter
- suffixes on memory text files are now respected
- if suffix exists and memory is aggregate, aggregate sub-fields will now be inserted before suffix
- every bind module created gets a unique number
- this is required when multiple loaded memories appear in a module
- this should be generalized for other uses of binding modules
* Support for LoadMemory annotation
- remove un-needed suffix test
* Support for LoadMemory annotation
- remove instance walk, now just processes each module
* Support for LoadMemory annotation
- Move LoadMemoryTransformation into Firrtl for treadle to access it.
* Support for LoadMemory annotation
- One more bug in suffix handling has been eliminated
* Support for LoadMemory annotation
- remove unused findModule per jackkoenig
- fixed complex test, bad filename edge case
* Support for LoadMemory annotation
- changed to not use intellij style column alignment for : declarations
* Load memory from file
Fixes based on @jkoenig review
- remove unused BindPrefixFactory
- Moved code from CreateBindableMemoryLoaders into to LoadMemoryTransfrom
- Made map to find relevant memory annotations faster
- Made map to find modules referenced by annotations faster
- Made things private that should be private
- DefAnnotatedMemorys are no longer referenced, shouldn't be found here.
- println of error changed to failed
* Loading memories from files
- Many changes based on review
- move stuff into experimental
- clean up annotation manipulation
- manage tests better
- use more standard practices for transform
* Loading memories from files
- More review changes
- Move doc from annotation to the object apply method that generates the annotation
- Make scalastyle directives more specific
- Use more efficient collect to generate name to module map
- Made lines obey style length limit
- a couple of cleanups of imports in tests
- removed some commented out code
- optimized checking for lines using .exists
- use _ for unused variable in match
|
|
This adds a new trait, FlattenInstance, to chisel3.util.experimental. When
mixed into a module or a specific instance this trait will "flatten",
i.e., "inline that module and all of its submodules".
This includes testing (additions to InlineSpec) and ScalaDoc
documentation.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
This adds a new trait, InlineInstance, to chisel3.util.experimental. This
trait, when mixed into a specific module or instance, will "inline" that
module, i.e., "collapse a module while preserving it's submodules."
This includes testing (InlineSpec) and ScalaDoc documentation.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
|
|
|
|
We can sometimes shim with other workarounds like VecInit or manually
creating a mux
|
|
|
|
|
|
|
|
This adds an annotator that provides a linkage to the FIRRTL
WiringTransform. This enables synthesizable cross module references
between one source and multiple sinks without changing IO (the
WiringTransform bores through the hierarchy).
Per WiringTransform, this will connect sources to their closest
sinks (as determined by BFS) or fail if ownership is indeterminate.
Make TesterDriver.execute work like Driver.execute:
- annotations are included when running FIRRTL
- custom transforms are run automatically
Also, add a bore method to BoringUtils that allows you to do one source to
multi-sink mapping in a single call. This adds a test that this is doing
the same thing as the equivalent call via disjoint addSink/addSource.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
|
|
They should not be deprecated until zero-width wires actually work
|
|
|
|
|
|
|
|
Documentation for example had parameters in wrong order, and was missing @param. Additionally, it was lacking a module wrapper. This has been corrected.
|
|
|
|
* Add test for UIntToOH
* Pad UIntToOH inputs to support oversized output widthds
* Optimize Bits.pad in case of known widths
* Add missing import and fix test in OneHotMuxSpec
|
|
* Generalize ChiselAnnotation
This allows us to delay creation of Annotations till elaboration is
complete. Also update all annotation-related code.
* Add RunFirrtlTransform
Use a Chisel-specific RunFirrtlTransform API to preserve behavior of old
ChiselAnnotation (now called ChiselLegacyAnnotation)
* Use unique test directories in ChiselRunners.compile
|
|
|
|
|
|
Add runtime warnings for use of deprecated Chisel methods. This is done using a macro that takes the message from a `@deprecated` annotation, and adds a call to `Builder.deprecated`.
Reasoning is that by default, Scala doesn't print all deprecations, and that it's somewhat tricky to notice them - yet some support questions revolve around the use of deprecated and terribad API. This now prints warnings for uses of deprecated functions at runtime, and aggregates them by error and line to avoid spam. Also included is convenient information on enabling scalac deprecations.
This also changes how line numbers for Chisel's error facility is determined, using prefix string comparison of the stack trace element classnames, instead of checking if the class is a subtype of UserModule. The previous one (specifically, calls to Class.forName) seems to interact badly with reflection-based cloneType when called at scale. This should also give more accurate reporting of errors that are in user code but outside of a UserModule.
It turns out that `@deprecated` on macro functions don't do anything, so this changes the tags to the functions that the macros point to, which seems to work properly. It also turns out that there's a bunch of uses of deprecated functions in chiselTests which needs to be fixed.
Not all `@deprecated` functions are also annotated with `@chiselRuntimeDeprecation`, because they're still used in Chisel internals, and we can't track whether they're called by the user or by Chisel and it will give a misleading error. These are a small amount of functions.
|
|
|
|
Also enforce switch is conditions are mutually exclusive literals
|
|
* Require explicit connection to DontCare to generate "is invalid".
* Add tests for RefNotInitializedException.
Currently, we fail the when ... otherwise ...
* Disable ScalaTest shrinking on error in ComplexAssignSpec.
* fix broken merge; still some binding issues
* cleanup DontCare connection checks; add missing directions to test module IOs
* Have library code inherit compileOptions from the enclosing Module (if it exists).
* work around current firrtl uninitialized references with Strict compile options and explicitInvalidate
* more CompileOptions cleanup; move test-specific defines to package object
* minimize differences with master
* set default CompileOptions.explicitInvalidate to false until we fix the FIRRTL when issue
* ignore the StrictCompiler property checks (until CompileOptions.explicitInvalidate is defaulted to true)
* Revert "more CompileOptions cleanup; move test-specific defines to package object"
This reverts commit e4486edcba990d150e76e08a2fc6abca033556e0.
* Revert "work around current firrtl uninitialized references with Strict compile options and explicitInvalidate"
This reverts commit 426faa430a62c3dac2dbdf33044d3386d4243157.
* remove unused code
* Convert to binding-based DontCare implementation
* comment cleanup to minimize differences with master
* Tentatively remove possibly redundant DefInvalid on module ports.
* Respond to code review change request.
- backout build.sbt change
- correct indentation
- handle bulk of DontCare semantics in elemConnect()
- have DontCare extend Element, not Data (eliminate most Object specific methods
- add comments indicating reason for explicit DontCare connections
* Initialize test elements without requiring a DontCare.
* Respond to review change requests.
- DontCare should work on left or right side in BiDirectional connections
- call bind() to set DontCare binding instead of messing with internal variables
- DontCares are only equivalent with DontCares
- clean up processWhens() definition
* Eliminate DontCare connection to inputs in MonoConnect().
* Pull aggregates apart for the purpose of DontCare connections.
* Restore the explicit (conditionally executed) ports DefInvalidin ImplicitModule()
* Don't add DontCare's to the module list of _ids.
* Add missing DefInvalid() to LegacyModule().
* Respond to review requests: add DontCare BiConnect Vec, remove null parent hack to avoid addId(), initialize singletons early in Builder
* Move DontCare out of chisel3.experimental.
|
|
Addresses #419
cloneType is now marked (through comments only) as an internal API.
chiselCloneType deprecated (and changed to cloneTypeFull internally, analogous to cloneTypeWidth).
chiselTypeOf(data) introduced as the external API to get a chisel type from a hardware object
Intended usage: cloning is an implementation detail, and chisel types and hardware objects both should act as immutable types, with operations like Input(...), Reg(...), etc returning a copy and leaving the original unchanged. Hence, the clone operations are all deprecated.
Deletes what appears to be an unused Bundle companion object.
Input(...), Output(...), Flipped(...) require the object to be unbound
|
|
|
|
Rest of the binding refactor
|
|
No functional changes
|
|
Part 1 of mega-change in #578
Major notes:
- Input(...) and Output(...) now (effectively) recursively override their elements' directions
- Nodes given userDirection (Input, Output, Flip - what the user assigned to _that_ node) and actualDirection (Input, Output, None, but also Bidirectional and BidirectionalFlip for mostly Aggregates), because of the above (since a higher-level Input(...) can override the locally specified user direction).
- DataMirror (node reflection APIs) added to chisel3.experimental. This provides ways to query the user given direction of a node as well as the actual direction.
- checkSynthesizable replaced with requireIsHardware and requireIsChiselType and made available in chisel3.experimental.
Internal changes notes:
- toType moved into Emitter, this makes the implementation cleaner especially considering that Vec types can't be flipped in FIRRTL. This also more clearly separates Chisel frontend from FIRRTL emission.
- Direction separated from Bindings, both are now fields in Data, and all nodes are given hierarchical directions (Aggregates may be Bidirectional). The actualDirection at the Element (leaf) level should be the same as binding directions previously.
- Bindings are hierarchical, children (of a, for example, Bundle) have a ChildBinding that points to their parent. This is different than the previous scheme where Bindings only applied at the Element (leaf) level.
- Lots of small misc clean up.
Future PRs will address other parts of #578, including stricter direction checks that aren't a side-effect of this internal refactor, stricter checks and splitting of binding operations (Wire vs. WireInit), and node operations not introduced here (getType and deprecation of chiselCloneType). Since those shouldn't mess with internals, those should be much smaller.
|
|
|
|
Replace ambiguous bi-connect ("<>") with mono-connect (":=") for internal Pipe wiring.
|
|
* Update comments describing Decoupled/ReadyValid.
It seems there is a valid use case for EnqIO/DeqIO and updating the comments may clear some of the confusion and encourage their usage.
* Update comments - no functional changes.
Re-flow comments for ReadyValidIO()
Add gen param to DecoupledIO() and IrrevocableIO().
* Update code and comment now that #492 is resolved
|
|
|
|
* Remove explicit import of NotStrict - fixes #492
* Provide macro for MemBase.apply().
* Provide macro for MemBase.apply().
Since a macro cannot override an abstract method, provide a concrete
apply method n VecLike() that we can override with a macro.
* Remove concrete apply() in VecLike.
Since MemBase no longer extends the trait VecLike, we do not require a concrete method to which we can apply a macro to extract the appropriate CompileOptions.
* Add missing implicit compileOptions to do_pad() and do_zext().
The latter caused:
```
[error] /vm/home/jenkins/workspace/rocket-chip_with_chisel3/hardfloat/src/main/scala/MulAddRecFN.scala:205: too many arguments for method do_zext: (implicit sourceInfo: chisel3.internal.sourceinfo.SourceInfo)chisel3.core.SInt
[error] val CDom_sExp = io.fromPreMul.sExpSum - io.fromPreMul.doSubMags.zext
```
* Add SourceInfoTransform macros to Vec methods in order to avoid apply() chain issues.
Since utils methods are no longer NotStrict, Pipe objects need access to the client's compile options. There may be more.
* Respond to review comments.
Don't propagate SourceInfo through helper functions.
Replace old usages of CompileOptionsTransform with the now equivalent SourceInfoTransform and redefine CompileOptionsTransform to only deal with CompileOptions.
Just thread CompileOptions (not SourceInfo) through deprecated functions.
|
|
* Remove explicit import of NotStrict - fixes #492
* Provide macro for MemBase.apply().
* Provide macro for MemBase.apply().
Since a macro cannot override an abstract method, provide a concrete
apply method n VecLike() that we can override with a macro.
* Remove concrete apply() in VecLike.
Since MemBase no longer extends the trait VecLike, we do not require a concrete method to which we can apply a macro to extract the appropriate CompileOptions.
|
|
|
|
Fixes #554
|
|
|
|
|
|
The old implementation failed to check for width <= -2, and
did the wrong thing when -1 was explicitly passed. Splitting
into two methods avoids the latter issue.
log2Ceil's argument might be 1, so employ a max operator.
|
|
Since the argument is at least 2, this change has no semantic effect.
|
|
|
