| Age | Commit message (Collapse) | Author |
|---|
|
* Copy MemConf.scala from ucb-bar/barstools#35 into memlib.
This provides a data structure wrapper around the existing memory conf format
which contains both reading and writing methods, making it easier to write code
that needs to read the format.
* Add MemConf tests and use a Map[MemPort, Int] for port lists instead of a Seq[MemPort] which is a bit less fragile.
|
|
* Don't create nodes to hold Muxes with >0 void cases
* Added test case demonstrating void error
* Memoize intermediate expression when checking for WVoid-ness
|
|
* Run CheckHighForm after all non-emitter transforms in firrtl tests
* Remove shlw from checks.scala
* Removed mistake in fix
* Fix FirrtlSpec fix
|
|
|
|
This fixes all Scaladoc warnings except for those trying to link to
Java.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
* refactor InferWidths to allow for extra contraints, add InferWidthsWithAnnos
* add test cases
* add ResolvedAnnotationPaths trait to InferWidthsWithAnnos
* remove println
* cleanup tests
* remove extraneous constraints
* use foreachStmt instead of mapStmt
* remove support for aggregates
* fold InferWidthsWithAnnos into InferWidths
* throw exception if ref not found, check for annos before AST walk
|
|
Fixes #219
* Adds AsyncResetType (similar to ClockType)
* Registers with reset signal of type AsyncResetType are async reset
registers
* Registers with async reset can only be reset to literal values
* Add initialization logic for async reset registers
|
|
|
|
This uses the foldShiftRight method of the ConstantPropagation
Transform when legalizing Shr PrimOps. This has the effect of removing
literals with bit extracts from the MinimumVerilogCompiler.
This makes the formerly private foldShiftRight method of a public
method of the ConstantPropagation companion object.
Tests in the MimimumVerilogCompilerSpec are updated to check that Shr
is handled as intended.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
* Add memory WRef factory for completeness
* Refactor DefAnnotatedMemory construction for clarity
|
|
* Improve Shl codegen; eliminate Shlw WIR node
If we emit shl(x, k) as {x, k'h0} instead of (x << k), then there's
no need for Verilog-specific padding in the PadWidths pass. Avoiding
the redundant padding improves compiler/simulator performance and
renders Shlw unnecessary.
* [skip formal checks] Add test
|
|
|
|
* Give better error when mport references non-existent memory
* Closes #796
|
|
|
|
* Added Foreachers
* Changed CheckTypes to use foreach
* Check widths now uses foreach
* Finished merge, added foreachers to added stmts
* Address reviewer feedback
|
|
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
This shows an example of using a RegisteredLibrary, with the appropriate
META-INF ServiceLoader entry, that adds options from the InferReadWrite
and ReplSeqMem transforms.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
This changes the CheckTypes.UniferredWidth exception to include the pretty
printed Target that was uninferred and suggests to the user that they may
have forgotten to assign to it. This changes the CheckTypes pass to
communicate the necessary Target information during AST traversal such
that when an uninferred width is found, the Target is known and available.
This also adds one test checking the message of the UniferredWidth
exception.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
|
|
It is O(n) and every use is in an O(n) iteration resulting in O(n^2).
Same information can be extracted from create_exps which happens to also
be called at every use of get_flip.
|
|
|
|
|
|
|
|
|
|
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.
|
|
This moves VerilogRename out of Passes.scala and genericizes it as the new
Transform KemoveKeywordCollisions. This new Transform will remove keywords
for arbitrary sets of reserved keyword.
This adds VerilogRename back as a class instead of an object.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Summary of changes to firrtl.passes.VerilogRename:
- Use "_" to mangle names that conflict with Verilog
keywords (previously "$")
- Rewrite to operate on the whole AST to propogate mangled ports and
module names
- Make VerilogRename a Transform (was previously a Pass)
- Renames are now propagated
- Adds documentation for new VerilogRename
This makes the VerilogRename Transform (previously a Pass) use an
underscore ('_') instead of a dollar sign ('$') to mangle names that
conflict with Verilog keywords. This prevents problems with potentially
buggy tools that are not expecting '$' in Verilog names.
This reimplements VerilogRename to be safe for name collisions that may
occur anywhere in the AST, e.g., in ports, module names, circuit names, or
in any statements/expressions. Previously, names were only mangled in
statements and in place. This resulted in problems where renames of ports
in a child's namespace would not be guaranteed to be mangled the same way
in a parent's namespace. The algorithm is reimplemented to walk all
modules in reverse topological order (from leafs to main) and relying on a
RenameMap to track name changes.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Summary of changes:
- Use "_" as an inlining delimiter instead of "$"
- Makes inlining avoid namespace conflicts
This changes the delimiter used for inlining to "_" instead of "$". This
avoids problems with buggy parsers that may not handle "$" correctly. As
ClockListTransform relies on the explicit use of "$", the delimiter is a
FIRRTL-private val that the ClockListTransform overrides (to the original
"$").
Namespace conflicts could occur previously, but are very rare as users
will almost never use "$" in a name (even though it's allowed by both the
FIRRTL and Verilog specifications). Moving to "_" increases the
probability of namespace conflicts occurring. This adds explicit checking
that inlined names will not introduce namespace conflicts and that
generated names are prefix unique (as defined in the spec).
Note: inlined modules may not have unique prefixes. A test is included
that this is the case and an ignored test shows what prefix uniqueness
would look like.
MISC:
- [skip chisel tests]: Changing the delimiter causes the Chisel
InlineSpec to fail as this explicitly checks for "$".
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
This makes findValidPrefix and enumerateNames both private to
FIRRTL (previously, these were private). This enables their use for name
generation by other FIRRTL passes/transforms.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Previously, mems marked no dedup would prevent mems with the same
instance name in other modules from deduping
|
|
* add FoldADD to const prop, add yosys miter tests
* add option for verilog compiler without optimizations
* rename FoldLogicalOp to FoldCommutativeOp
* add GetNamespace and RenameModules, GetNamespace stores namespace as a ModuleNamespaceAnnotation
* add constant propagation for Tail DoPrims
* add scaladocs for MinimumLowFirrtlOptimization and yosysExpectFalure/Success, add constant propagation for Head DoPrim
* add legalize pass to MinimumLowFirrtlOptimizations, use constPropBitExtract in legalize pass
|
|
Previously, Vecs of Bundles that contained a zero-width element would
result in a ClassCastException
|
|
On circuits with large numbers of width inferences, prepend to a linked
list instead of appending and having to make a copy.
Fixes #842
|
|
[skip formal checks] LEC passes with Formality
* Improve code generation for smem RW-port wmode port
A common case for these port-enables is
wen = valid & write
ren = valid & !write
which the RW-port transform currently turns into
en = (valid & write) | (valid & !write)
wmode = valid & write
because it proved `wen` and `ren` are mutually exclusive via `write`.
Synthesis tools can trivially optimize `en` to `valid`, so that's not a
problem, but the wmode field can't be optimized if going into a black box.
This PR instead sets `wmode` to whatever node was used to prove
mutual exclusion, which is always a simpler expression. In this case:
en = (valid & write) | (valid & !write)
wmode = write
* In RemoveCHIRRTL, infer mask relative to port definition
Previously, it was inferred relative to the memory definition causing
the mask condition to be redundantly conjoined with the enable signal.
Also enable ReplSeqMems to ignore all ValidIfs (not just on Clocks) to
improve QoR.
|
|
Fixes #756
|
|
Also minor cleanup to literal construction in Visitor
|
|
This enables the pattern of attaching "through" a wire to give better
Verilog that also works in Verilator
Use WrappedExpression when combining attaches in ExpandWhens
to ensure no duplication of references in resulting, combined attaches
|
|
* Collects Infos found for symbols
* Merges multiple sources for symbol into MultiInfo
* Restores these Infos on connect statements.
* Add test showing preserved Infos
* Changed ++ methods on the Info sub-classes
* Ignore NoInfo being added
* Fixed way adding was implemented in MultiInfo
* Made InfoMap a class which defines the default value function
|
|
This should close #757. It should also allow for stop() and printf()
to be used with zero-width fields.
|
|
Replaces old VerilogWrap which didn't work with split expressions and was
actually buggy anyway. This functionality reduces unnecessary intermediates in
emitted Verilog.
|
|
The following are deprecated in favor of DiGraph/InstanceGraph:
- firrtl.passes.wiring.Lineage
- firrtl.passes.wiring.WiringUtils.ChildrenMap
- firrtl.passes.wiring.WiringUtils.getChildrenMap
- firrtl.passes.wiring.WiringUtils.getLineage
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
* Change VerilogMemDelays to put new Statements at end of Module
Fixes #547
This is instead of putting them right after the modified DefMemory which could
result in use before declaration errors for things that feed into the new
logic.
* Adds tests that show VerilogMemDelays crashing. (#792)
|
