| Age | Commit message (Collapse) | Author |
|
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
* Add test to check ShiftRegister(s) with delay is 0.
This should break ShiftRegister(x, 0) since last is not exist in a empty
Seq. Originally, test only test 1 to 4, which missed a potential bug
from #1723.
* Fix ShiftRegister with 0 delay.
if ShiftRegisters is empty, java will complain:
```
java.util.NoSuchElementException
scala.collection.LinearSeqOptimized.last(LinearSeqOptimized.scala:150)
```
This fix this issue and return `in` directly when ShiftRegister size is 0.
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
* add ShiftRegisters to expose register inside ShiftRegister.
* use Seq.iter for oneline implementation.
|
|
|
|
* add convert(chirrtl: cir.Circuit): fir.Circuit to convert chirrtl to firrtl.
* add scaladoc.
* add test.
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
|
|
|
|
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
Change top.cpp to deassert reset one time unit before the clock asserts.
This avoids a Verilator simultation issue in top.cpp where the eval()
function is only called once per simultation loop. If the clock and
reset are both changed and eval() is only called once, then any
combinational update due to a change in reset is not visible to the
sequential logic. This avoids issues where the downstream compilation
utilities move synchronous reset logic outside of an always block that
describes a synchronous reset flip flop.
Reset now deasserts on time unit 10 and the clock ticks on time unit
11.
h/t @albert-magyar
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@sifive.com>
|
|
This annotation adds memory import with inline generation for the
emmiter.
Supports both readmemh and readmemb statements based on argument.
|
|
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
|
|
* Handle MemTypeBinding in Analog
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@sifive.com>
* Fix stack trace trimming across ChiselStage
Fix bug in stack trace trimming behavior. Now, the following is what
happens:
1. The Builder, if catching accumulated errors, will now throw a
ChiselException with a Scala-trimmed Stack trace. Previously, this
would throw the full excpetion.
2. The Elaborate phase handles stack trace trimming. By default, any
Throwable thrown during elaboration will have its stack
trace *mutably* trimmed and is rethrown. A logger.error is printed
stating that there was an error during elaboration and how the user
can turn on the full stack trace. If the --full-stacktrace option
is on, then the Throwable is not caught and only the first
logger.error (saying that elaboration failed) will be printed.
3. ChiselStage (the class), ChiselStage$ (the object), and ChiselMain
all inherit the behavior of (2).
Mutable stack trace trimming behavior is moved into an implicit
class (previously this was defined on ChiselException only) so this
can be applied to any Throwable.
No StageErrors are now thrown anymore. However, StageErrors may still
be caught by ChiselMain (since it is a StageMain).
Testing is added for ChiselMain, ChiselStage, and ChiselStage$ to test
all this behavior.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@sifive.com>
|
|
* Added SyncReadMem-based implementation of the Queue class
* Rework of the parametrized Queue class SyncReadMem-based implementation
* Modification of the parametrized Queue class SyncReadMem-based implementation
* Limiting the visibility of the read address for SyncReadMem-based Queue
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
|
|
If a method passed to higher function does not return any value,
it is prefer to use `foreach` instead of `map`.
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
|
|
Chisel projects no longer need -Xsource:2.11 when compiling with Scala
2.12.
Autowrapping of "val io" for compatibility mode Modules is now
implemented using reflection instead of calling the virtual method.
Also move Chisel.BlackBox to new chisel3.internal.LegacyBlackBox
|
|
|
|
* Improve source locators for switch statements.
|
|
|
|
* Bugfix - module name collision for injecting aspect
* Fixed mechanism to avoid module name collisions
* Added comments for reviewer feedback
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
* Added forcename transform and tests
* Added documentation and additional error checking
* Added mdoc. Added RunFirrtlTransform trait
* Removed TODO comment
* Addressed reviewer feedback
* Removed trailing comma
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
* Fix `-e` option causing ChiselStage.emit* to error
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@sifive.com>
* Add test of `-e` ChiselStage behavior
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@sifive.com>
* fixup! Add test of `-e` ChiselStage behavior
|
|
* Return 0.U for asUInt of a zero-element Seq
Add a condition to SeqUtils.asUInt to have it return an unspecified
width 0.U when applied to an empty sequence. This enables the ability
to do a Cat of a zero-element sequence.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@sifive.com>
* Test elaboration of Cat on zero-element Seq
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@sifive.com>
|
|
* ExtModule's lacked support built in support for providing
the verilog source. This changes creates traits that
can be used with ExtModule to provide the support currently found in
BlackBox
- Add support for ExtModule helpers
- HasExtModuleResource to use addResource
- HasExtModuleInline to use setInline
- HasExtModulePath to use addPath
- Add tests of the above support.
- Note: These tests use Stage instead of Driver
- Added ScalaDoc for HasBlackBoxInline#setInline
* Fix the danged trailing commas.
* Change to use `.transform` as the correct API for `ChiselStage`
|
|
Change source and other relevant files to use SPDX license
LICENSE file moved from src/ to ./
Changed license file to refer to this per recommendation
using_spdx_license_list_short_identifiers
WARNING: Tests fail with as of yet undiagnosed error
```
[error] Failed: Total 691, Failed 19, Errors 0, Passed 672, Ignored 15
[error] Failed tests:
[error] chiselTests.QueueSpec
[error] examples.VendingMachineGeneratorSpec
[error] chiselTests.HarnessSpec
[error] chiselTests.ConnectSpec
[error] chiselTests.aop.SelectSpec
[error] chiselTests.PopCountSpec
[error] chiselTests.CloneModuleSpec
[error] (Test / test) sbt.TestsFailedException: Tests unsuccessful
[error] Total time: 379 s (06:19), completed Sep 30, 2020 12:38:17 AM
sbt:chisel3>
```
|
|
|
|
Co-authored-by: Megan Wachs <megan@sifive.com>
|
|
ChiselCircuitAnnotation no longer extends CustomFileEmission, rather it
is Unserializable. Also the --chisel-output-file is added to the
ChiselCli.
New phase AddSerializationAnnotations constructs a
CircuitSerializationAnnotation from ChiselCircuitAnnotation and
ChiselOutputFileAnnotation. Both .fir and .pb file formats are
supported. Default format is .fir unless a --chisel-output-file is
specified with a .pb extension.
|
|
* fix loadMemoryFromFile to work with binary
Passed in hexOrBinary parameter to ChiselLoadMemoryAnnotation
* Added test for binary format support in loadMemoryFromFile
* Added test for binary format support in loadMemoryFromFile
|
|
It shouldn't be removed until we also remove the underlying chisel3.Driver.
|
|
ChiselStage$ helpers (#1566)
* Add ChiselPhase
* Use ChiselPhase in ChiselStage, remove targets
Switch from a one-off PhaseManager inside ChiselStage to actually
using the newly added ChiselPhase. This removes the targets
method (and API) from ChiselStage.
* Stop writing to files in ChiselStage$ methods
Change the ChiselStage companion object methods, elaborate and
convert, to not write files. Under the hood, these are switched from
using ChiselStage (which, like all phases, will write files) to using
ChiselPhase.
* Test that ChiselStage$ methods write no files
Modify existing ChiselStage object method tests to check that no files
are written.
* Expand ChiselStage$ API with more helpers
This adds additional methods to the ChiselStage object for going
directly from a Chisel module to a string including: CHIRRTL, high
FIRRTL IR, Verilog, and SystemVerilog.
Differing from their ChiselStage class counterparts, these take no
arguments other than the module and write no files.
* Add tests of new ChiselStage$ helper methods
* Use ChiselStage object in tests
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Remove usages of the deprecated trait PreservesAll and use an explicit
false invalidates. Additionally, all phases are converted to be more
canonical in there specification of dependencies by:
1. Overriding all default dependency implementations
2. Using def instead of val
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Use inheritance to make TesterDriver Backend API extensible, then define
a TreadleBackend in the test project
|
|
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
* Fix emit{Firrtl,Verilog} for CustomFileEmission
Change ChiselStage helper methods for emitting FIRRTL (emitFirrtl) and
Verilog (emitVerilog) to look for Circuit and Verilog annotations
instead of DeletedAnnotations. This is needed after migrating to the
CustomFileEmission mixin in FIRRTL where FIRRTL will no longer delete
emitter annotations.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
* Use CustomFileEmission for ChiselCircuitAnnotation
Removes the explicit chisel3.phases.Emitter and instead does emission
with a CustomFileEmission mixin to ChiselCircuitAnnotation. This then
prevents the need for passing around DeletedAnnotations. As a
consequence, I removed an unnecessary run of a second Converter in the
Driver.
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
* Fix tests for use of CustomFileEmission trait
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
* Fixes for newer CustomFileEmission API
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|
|
Includes special case support for Counter(0) which has identical
behavior to Counter(1) except for the value of n.
|
|
comment fix only.
|
|
|
|
* Add positive range generator
* Allow the Counter module to be instantiated with a Scala range
* Use head/last to determine counter width
Co-authored-by: Jack Koenig <jack.koenig3@gmail.com>
* Let counter overflow naturally when appropriate
We only need to explicitly wrap counters that don't start at zero, or end on a power of two. Otherwise we just let the counter overflow naturally to avoid wasting an extra mux.
* Require counter range to be non-empty
Co-authored-by: Jack Koenig <jack.koenig3@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
Added prefixing and a compiler plugin to improve naming. Only works for Scala 2.12 and above.
Co-authored-by: Jack Koenig <koenig@sifive.com>
|
|
|
|
Treadle tester phase requires new to fix problems found
when running tests in parallel.
|
|
* Add `check(...)` affordance
* Add assert (renamed from check and fixed)
* Add verification statements
* Move formal to experimental.verification
* Make test use ChiselStage
`generateFirrtl` has been cut from Chisel
* Fix newly introduced style warnings
* Fix some old style warnings for good measure
* Revert "Fix some old style warnings for good measure"
This reverts commit 31d51726c2faa4c277230104bd469ff7ffefc890.
* Cut scalastyle comments
* Cut formal delimiter comments
|
|
|
|
|
|
This mechanism is not enabled and should not change the behavior of existing tests
A following PR will deliver a switch that will allow changing the backend.
The reasons for this PR
- Treadle tests run much faster, enabling quicker debugging and CI cycles
- This will help ensure fidelity of Treadle to the Verilator backend
A few tests are marked as verilator only due to black box limitations
Change treadle to a direct dependency
I tried to make it a test only dependency but the TesterDriver sits in src/main requiring that
regular compile have access to treadle
Oops, made treadle the default
A number of changes in response to @ducky64 review
- made backend check clearer and add error handling for multiple backends specified
- Fixed duplicate TargetDirAnnotation uses in Treadle backend
- Cleaned up BlackBox test formatting
- Undid unnecessary debugging changes from Counter
- Undid .gitignore change, that should be on another PR
A number of changes in response to @ducky64 review
- Undid debugging changes made to BitWiseOps
|
|
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
|
|
Co-authored-by: Chick Markley <chick@qrhino.com>
Co-authored-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
Signed-off-by: Schuyler Eldridge <schuyler.eldridge@ibm.com>
|