| Age | Commit message (Collapse) | Author |
|---|
|
* Add renamemap.MutableRenameMap which includes these methods without
deprecation
* Deprecate Stringly typed RenameMap APIs which were accidentally
undeprecated a while ago
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
* Add transform to deduplicate memory annotations
* Add annotation deduplication to Dedup stage
* ResolveAnnotationPaths and EliminateTargetPaths now invalidate the dedup annotations transform
* Verilog emitter now throws exception when memory annotations fail to dedup
Co-authored-by: Jack Koenig <koenig@sifive.com>
|
|
|
|
|
|
* InstanceKeyGraph: add staticInstanceCount, getGraph and getChildrenInstanceMap
* InstanceKeyGraph: reachableModules, unreachableModules, lowestCommonAncestor and fullHierarchy
* Replace usage of InstanceGraph with InstanceKeyGraph
Also deprecates all unused methods.
* WiringUtils: make new version of sinksToSources package private
This will make our live easier next time we need to change it.
* CircuitGraph: use InstanceKeyGraph
* InstanceKeyGraphSpec: respect maximum line width
* InstanceKeyGraph: make constructor private
* InstanceKeyGraph: move lowestCommonAncestor function to Wiring
* WiringUtils: update deprecation message
|
|
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
- RenameMap Behavior
-- Prevent transitive renaming A -> B -> C (continueRenaming)
-- Prevent transitive renaming for self-renames
- Target
-- Override toString as serialize for CompleteTarget
-- Expansion of stripHierarchy to enable stripping InstanceTargets to become ModuleTargets
Annotations
-- Bugfix in extractComponents where Products were not iterated over
-- Converts renamed targets to local targets using Target.referringModule to preserve sticky behavior
- Eliminate Target Paths
-- Make DuplicationHelper use LinkedHashMap, as we iterate over its contents and convert to Seq in def makePathless
-- Add DupedResult to map original module to new module targets
-- Update renaming to record a map from all relative instance paths to original module, to new module target
-- Consumes DedupedResult to give better name to new duplicated module if it was originally deduplicated
-- Reorder modules in attempt to preserve original ordering, pre-deduplication
-- Move utility functions to object
-- Bugfix: add self-renames to prevent ofModule _ of target _ cannot be renamed to Vector(_, _, _, ...) errors
- Dedup
-- Changed NoDedupAnnotation to contain ModuleTarget, rather than ModuleName
-- Added DedupedResult to map original module to the duplicate module
-- Consumes DupedResult to pick better name, if it existed
-- Updates renaming to chain the following: instancify deduped modules, remap differently named internal signals, then remap AST modules
-- Move utility functions to object
-- Remove annotations as part of determination of dedup correctness
-- Bugfix: add instance renames so that deduped modules have their instances properly renamed
- Dead Code Elimination
-- Add deletion of ASTModules
- Tests
-- Morphism Spec to ensure Dedup -> EliminateTargetPaths and EliminateTargetPaths -> Dedup patterns work properly
-- Update existing tests to make sure they work properly
-- Add Dedup tests to demonstrate instance renaming bug, EliminateTargetPaths for ofModule rename bug, and update RenameMap tests
Co-authored-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
Co-authored-by: Adam Izraelevitz <adam.izraelevitz@sifive.com>
Co-authored-by: Adam Izraelevitz <azidar@gmail.com>
Co-authored-by: Jack Koenig <koenig@sifive.com>
|
|
* Absorb WRef into Reference
* Absorb WSubField into SubField
* Absorb WSubIndex into SubIndex
* Absorb WSubAccess into SubAccess
* Absorb WDefInstance into DefInstance
------------------------- API CHANGE SEVERITY --------------------------
This is projected to not break source-level compatibility with any known
user code. However, it will break *binary* compatibility with all
existing user FIRRTL passes, as is generally allowed with major
releases of FIRRTL.
--------------------------- DESCRIPTION --------------------------------
Previously, there were several nodes in WIR.scala that had a one-to-one
correspondance with existing nodes in the standard firrtl.ir hierarchy.
These nodes would have a case class resembling the corresponding
standard IR node, but with the addition of one or more "analysis"
fields.
Since these fields (such as kind) represent helpful info that can be
invalidated or set to Unknown (e.g. UnknownKind for Kind), it does not
cause any issues to simply include these fields on any in-memory
representation of FIRRTL IR. Although other systems for tracking FIRRTL
analyses have evolved over time, the ubiquity of pattern-matching on
these fields has lead most core and custom transforms to be written
against WIR, rather than IR.
This PR unifies the IRs by adding the fields that would be in an
"augmented" WIR node directly into the corresponding IR node; i.e., the
"type" and "kind" fields from WRef are added directly to the definition
of the Reference case class, while these "repetitive" WIR case classes
are removed entirely.
-------------------- SOURCE-COMPATIBILITY ADAPTERS ---------------------
Several object methods are added to WIR.scala to maintain
source-compatiblity for passes that used WIR. These objects define
factory methods and unapply methods, so passes that relied on implicit
case class factories or pattern matching for the removed WIR types will
remain perfectly source-compatible. However, these do not guarantee
compatibility at the binary level.
The types of the removed WIR case classes are also added as type aliases
to the top-level firrtl package, which allows code that relies on
explicit constructor calls or reflection to retain source-compatibility.
Finally, additional explicit factory methods are added to the companion
objects of the newly-augmented IR case classes, which allows user code
to avoid having to specify any of the new analysis fields. Existing code
that created non-WIR IR nodes will be able to continue using the
previous factory signatures, which will cause all omitted analysis
fields to be set to Unknown.
---------------------- UNMITIGATED API CHANGES -------------------------
While passes that used WIR will be source-compatible with this change,
there is one significant change that affects any pass currently using
non-WIR IR: the signatures of pattern-matching cases for Reference,
SubField, SubIndex, SubAccess, and DefInstance must change to
accommodate the extra fields.
This cannot be worked at the API level due to restrictions on unapply
overloading, but it could theoretically be solved with macros or other
static rewriting. However, only four core transforms (RemoveProto,
ToWorkingIR, Dedup, and RemoveChirrtl) use non-WIR IR, and it is
expected that no user code currently relies on it, so the expected
migration strategy is simply to change the small fraction of code
relying on these nodes.
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
This mixes in the new DependencyAPIMigration trait into all Transforms
and Passes. This enables in-tree transforms/passes to build without
deprecation warnings associated with the deprecated CircuitForm.
As a consequence of this, every Transform now has UnknownForm as both
its inputForm and outputForm. This PR modifies legacy Compiler and
testing infrastructure to schedule transforms NOT using
mergeTransforms/getLoweringTransforms (which rely on inputForm and
outputForm not being UnknownForm), but instead using the Dependency
API.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
* EliminateTargetPaths: add lone instance test cases
* EliminateTargetPaths: don't rename lone instances
* get rid of trailing comma
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
Co-authored-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
Co-authored-by: Albert Magyar <albert.magyar@gmail.com>
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Change the behavior of EliminateTargetPaths to generate ModuleTarget
renames when instances are duplicated. Previously, only InstanceTarget
renames would be generated.
In effect, annotations targeting a duplicated module when be
duplicated to point at the original and duplicated module.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Change EliminateTargetPaths to remove ResolvePaths annotations in the
output AnnotationSeq. This prevents a bug whereby the upstream
ResolvePaths annotations from previous runs of EliminateTargetPaths
can result in unexpected duplication.
Adds a test that checks that ResolvePaths annotations are actually
removed.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
|
|
Formerly #865
Major Code Changes/Features Added:
Added Target trait as replacement for Named
Added TargetToken as token in building Target
Added GenericTarget as a catch-all Target
Added CircuitTarget, ModuleTarget, ReferenceTarget, and InstanceTarget
Added ResolvePaths annotation
Added EliminateTargetPaths (and helper class DuplicationHelper)
Updated Dedup to work with instance annotations
Updated RenameMap to work with instance annotations
DCE & ConstantProp extend ResolveAnnotationPaths
|
|
This reverts commit 7e2f787e125227dc389d5cf1d09717748ecfed2e.
|
|
Added Target, which now supports Instance Annotations. See #865 for details.
|
