Vamsi K Vytla [Mon, 27 Apr 2020 16:44:24 +0000 (09:44 -0700)]
Preserve 'signed'-ness of a verilog wire through RTLIL
As per suggestion made in https://github.com/YosysHQ/yosys/pull/1987, now:
RTLIL::wire holds an is_signed field.
This is exported in JSON backend
This is exported via dump_rtlil command
This is read in via ilang_parser
Claire Wolf [Fri, 24 Apr 2020 12:09:47 +0000 (14:09 +0200)]
Merge pull request #1995 from YosysHQ/eddie/fix_verific_wiretype
verific: do not assert if wire not found; warn instead
Dan Ravensloft [Thu, 23 Apr 2020 21:44:29 +0000 (22:44 +0100)]
intel_alm: cleanup duplication
Eddie Hung [Thu, 23 Apr 2020 23:28:11 +0000 (16:28 -0700)]
verific: do not assert if wire not found; warn instead
Eddie Hung [Thu, 23 Apr 2020 13:43:30 +0000 (06:43 -0700)]
Merge pull request #1974 from YosysHQ/eddie/abc9_disable_mfs
abc9: tolerate &mfs failure by writing output file before calling it (and using that if it fails)
Claire Wolf [Thu, 23 Apr 2020 09:34:19 +0000 (11:34 +0200)]
Merge pull request #1989 from boqwxp/qbfsat_anyconst_sourcelocs
qbfsat: Make hole name recovery from source locations more robust.
Claire Wolf [Thu, 23 Apr 2020 09:33:54 +0000 (11:33 +0200)]
Merge pull request #1988 from boqwxp/qbfsat
qbfsat: Add `-assume-negative-polarity` option.
Claire Wolf [Thu, 23 Apr 2020 09:28:05 +0000 (11:28 +0200)]
Merge pull request #1986 from YosysHQ/eddie/verific_enum
verific: import enum attributes from verific
Dan Ravensloft [Wed, 22 Apr 2020 23:56:49 +0000 (00:56 +0100)]
intel_alm: work around a Quartus ICE
Alberto Gonzalez [Thu, 23 Apr 2020 05:42:18 +0000 (05:42 +0000)]
qbfsat: Make hole name recovery more robust. Allow multiple cell types to share the same source location as long as only one `$anyconst` or `$anyseq` has that location.
Eddie Hung [Thu, 23 Apr 2020 05:12:41 +0000 (22:12 -0700)]
Merge pull request #1984 from YosysHQ/eddie/getParam_exception
kernel: Cell::getParam() to throw exception again if not found
Alberto Gonzalez [Thu, 23 Apr 2020 04:06:15 +0000 (04:06 +0000)]
qbfsat: Add `-assume-negative-polarity` option.
Eddie Hung [Thu, 23 Apr 2020 00:53:08 +0000 (17:53 -0700)]
ecp5: ecp5_gsr to skip cells that don't have GSR parameter again
Eddie Hung [Thu, 23 Apr 2020 00:50:30 +0000 (17:50 -0700)]
tests: read +/xilinx/cell_sim.v before xilinx_dsp test
Eddie Hung [Thu, 23 Apr 2020 00:43:25 +0000 (17:43 -0700)]
xilinx: xilinx_dsp_cascade to check CREG for DSP48E1 only
Eddie Hung [Thu, 23 Apr 2020 00:26:56 +0000 (17:26 -0700)]
verific: import enum attributes from verific
Eddie Hung [Wed, 22 Apr 2020 23:35:35 +0000 (16:35 -0700)]
test: ice40_dsp test to read +/ice40/cells_sim.v for default params
Eddie Hung [Wed, 22 Apr 2020 20:59:56 +0000 (13:59 -0700)]
xilinx: improve xilinx_dffopt message
Eddie Hung [Wed, 22 Apr 2020 20:57:09 +0000 (13:57 -0700)]
xilinx: xilinx_dffopt to read cells_sim.v; fix test
Eddie Hung [Wed, 22 Apr 2020 19:07:08 +0000 (12:07 -0700)]
kernel: Cell::getParam() to throw exception again if not found
As it did before #1945
Eddie Hung [Wed, 22 Apr 2020 22:35:05 +0000 (15:35 -0700)]
Merge pull request #1949 from YosysHQ/eddie/select_blackbox
select: do not select inside black-/white- boxes unless '=' prefix used
Eddie Hung [Wed, 22 Apr 2020 21:37:27 +0000 (14:37 -0700)]
Merge pull request #1983 from YosysHQ/eddie/use_default_param
Cleanup use of hard-coded default parameters in light of #1945
whitequark [Wed, 22 Apr 2020 20:29:08 +0000 (20:29 +0000)]
Merge pull request #1982 from AsuMagic/asu/cxxrtl-memory-queue-opt
cxxrtl: keep the memory write queue sorted on insertion.
Claire Wolf [Wed, 22 Apr 2020 19:31:32 +0000 (21:31 +0200)]
Update passes/cmds/select.cc
Co-Authored-By: Eddie Hung <eddie@fpgeh.com>
Eddie Hung [Wed, 22 Apr 2020 19:10:42 +0000 (12:10 -0700)]
Merge pull request #1969 from boqwxp/pool_emplace
kernel: Add `pool` support for rvalue references and C++11 move semantics.
Eddie Hung [Wed, 22 Apr 2020 19:02:30 +0000 (12:02 -0700)]
Cleanup use of hard-coded default parameters in light of #1945
Asu [Wed, 22 Apr 2020 18:50:13 +0000 (20:50 +0200)]
cxxrtl: keep the memory write queue sorted on insertion.
Strategically inserting the pending memory write in memory::update to keep the
queue sorted allows us to skip the queue sort in memory::commit.
The Minerva SRAM SoC runs ~7% faster as a result.
Eddie Hung [Wed, 22 Apr 2020 17:19:30 +0000 (10:19 -0700)]
Merge pull request #1973 from YosysHQ/eddie/fix1966
tests: fix various/plugin.sh when PREFIX != /usr/local/share
Eddie Hung [Wed, 22 Apr 2020 17:16:14 +0000 (10:16 -0700)]
tests: update select black/white-box tests
Eddie Hung [Wed, 22 Apr 2020 17:15:56 +0000 (10:15 -0700)]
select: do not select black/white boxes by default, '=' prefix to do so
Eddie Hung [Wed, 22 Apr 2020 16:32:13 +0000 (09:32 -0700)]
Merge pull request #1950 from YosysHQ/eddie/design_import
design: -import to not count black/white-boxes as candidates for top
Eddie Hung [Wed, 22 Apr 2020 15:30:03 +0000 (08:30 -0700)]
yosys-config: spelling
Eddie Hung [Wed, 22 Apr 2020 15:29:45 +0000 (08:29 -0700)]
tests: use `yosys-config --datdir` instead of hard-coded
Eddie Hung [Wed, 22 Apr 2020 15:14:07 +0000 (08:14 -0700)]
pool: add emplace() function
Claire Wolf [Wed, 22 Apr 2020 14:57:34 +0000 (16:57 +0200)]
Merge pull request #1976 from YosysHQ/dave/fix-sim-const
sim: Fix handling of constant-connected cell inputs at startup
Claire Wolf [Wed, 22 Apr 2020 14:50:45 +0000 (16:50 +0200)]
Merge pull request #1979 from whitequark/cxxrtl-go-faster
cxxrtl: Gas gas gas! I'm gonna step on the gas! Tonight I'll fly!
whitequark [Wed, 22 Apr 2020 12:45:19 +0000 (12:45 +0000)]
cxxrtl: run edge detectors only once in eval().
As a result, Minerva SRAM SoC runs ~15% faster.
whitequark [Wed, 22 Apr 2020 01:15:27 +0000 (01:15 +0000)]
cxxrtl: add an unsupported knob for manipulating clock trees.
This is quite possibly the worst way to implement this, but it does
work for a subset of well-behaved designs, and can be used to measure
how much performance is lost simulating the inactive edge of a clock.
It should be replaced with a clock tree analyzer generating safe
code once it is clear how should such a thing look like.
whitequark [Tue, 21 Apr 2020 23:42:56 +0000 (23:42 +0000)]
cxxrtl: use log_id() where appropriate. NFC.
Marcelina Kościelnicka [Tue, 21 Apr 2020 17:56:28 +0000 (19:56 +0200)]
bugpoint: Don't remove modules or cells while iterating over them.
Reported by @ZirconiumX.
whitequark [Tue, 21 Apr 2020 21:48:17 +0000 (21:48 +0000)]
cxxrtl: add (*cxxrtl.{comb,sync}*) annotations on black box outputs.
If the annotations are not used, this commit does not alter semantics
at all, other than removing elision of outputs of black box cells.
(Elision of such outputs is expected to be too rare to have any
noticeable benefit, and the implementation was somewhat of a hack.)
The (* cxxrtl.comb *) annotation alters the semantics of the output
of the black box it is applied to such that, if the black box
converges immediately, no additional delta cycle is necessary to
propagate the computed combinatorial value upwards in hierarchy.
The (* cxxrtl.sync *) annotation alters the semantics of the output
of the black box it is applied to such as to remove any uses of
the black box by the wires connected to this output, and break false
feedback arcs arising from conservative modeling of dependencies of
the black box.
Although currently these attributes are only recognized on black
boxes, if separate compilation is added in the future, it could also
emit and consume them.
whitequark [Tue, 21 Apr 2020 18:46:36 +0000 (18:46 +0000)]
cxxrtl: s/sync_{wire,type}/edge_{wire,type}/. NFC.
The attribute for this is called (* cxxrtl.edge *), and there is
a planned attribute (* cxxrtl.sync *) that would cause blackbox
cell outputs to be added to sync defs rather than comb defs.
Rename the edge detector related stuff to avoid confusion.
Dan Ravensloft [Tue, 21 Apr 2020 15:43:21 +0000 (16:43 +0100)]
intel_alm: Documentation improvements
Alberto Gonzalez [Tue, 21 Apr 2020 17:17:47 +0000 (17:17 +0000)]
kernel: Rename arguments to rvalue-reference-accepting functions.
Marcelina Kościelnicka [Thu, 16 Apr 2020 13:57:03 +0000 (15:57 +0200)]
write_json: dump default parameter values
Fixes #1823.
This will allow nextpnr to reuse the default value information already
present in yosys cells_sim.v and avoid duplicating (and probably
desyncing) this information.
Marcelina Kościelnicka [Thu, 16 Apr 2020 19:48:21 +0000 (21:48 +0200)]
Use default parameter value in getParam
Fixes #1822.
Marcelina Kościelnicka [Thu, 16 Apr 2020 15:38:55 +0000 (17:38 +0200)]
hierarchy: Convert positional parameters to named.
Fixes #1821.
Marcelina Kościelnicka [Thu, 16 Apr 2020 13:51:03 +0000 (15:51 +0200)]
ilang, ast: Store parameter order and default value information.
Fixes #1819, #1820.
Marcelina Kościelnicka [Thu, 16 Apr 2020 19:48:03 +0000 (21:48 +0200)]
idict: Make iterator go forward.
Previously, iterating over an idict returned its contents in reverse.
Claire Wolf [Tue, 21 Apr 2020 17:04:55 +0000 (19:04 +0200)]
Merge pull request #1971 from YosysHQ/claire/edifkeep
Ignore conflicting keep attributes, unless asked not to
Claire Wolf [Tue, 21 Apr 2020 16:46:52 +0000 (18:46 +0200)]
Merge pull request #1851 from YosysHQ/claire/bitselwrite
Improved rewrite code for writing to bit slice
whitequark [Tue, 21 Apr 2020 15:51:09 +0000 (15:51 +0000)]
cxxrtl: use one delta cycle for immediately converging netlists.
If it is statically known that eval() will converge in one delta
cycle (that is, the second commit() will always return `false`)
because the design contains no feedback or buffered wires, then
there is no need to run the second delta cycle at all.
After this commit, the case where eval() always converges immediately
is detected and the second delta cycle is omitted. As a result,
Minerva SRAM SoC runs ~25% faster.
whitequark [Tue, 21 Apr 2020 15:33:12 +0000 (15:33 +0000)]
cxxrtl: add -O6, a shortcut for running `proc; flatten`.
People judge a compiler backend by the first impression, and
the metric they judge it for is speed. -O6 does severely impact
debuggability, but it provides equally massive gains in performance,
so use it by default.
whitequark [Tue, 21 Apr 2020 14:49:36 +0000 (14:49 +0000)]
cxxrtl: unbuffer module input wires.
Module input wires are never set by the module, so it is unnecessary
to buffer them. Although important for all inputs, this is especially
critical for clocks, since after this commit, hierarchy levels no
longer add delta cycles. As a result, Minerva SRAM SoC runs ~73%
faster when flattened, and ~264% (!!) faster when hierarchical.
whitequark [Tue, 21 Apr 2020 13:59:42 +0000 (13:59 +0000)]
cxxrtl: simplify generated edge detection logic.
This commit changes the way edge detectors are represented in
generated code from a variable that is set in commit() and reset in
eval() to a function that considers .curr and .next of the clock
wire. Behavior remains the same. Besides being simpler to generate
and providing more opportunities for optimization, this commit paves
way for unbuffering module inputs.
whitequark [Tue, 21 Apr 2020 13:33:42 +0000 (13:33 +0000)]
cxxrtl: localize wires with multiple comb drivers, too.
Before this commit, any wire that was not driven by an output port of
exactly one comb cell would not be localized, even if there were no
feedback arcs through that wire. This would cause the wire to become
buffered and require (often quite a few) extraneous delta cycles
during evaluation. To alleviate this problem, -O5 was running
`splitnets -driver`.
However, this solution was mistaken. Because `splitnets -driver`
followed by `opt_clean -purge` would produce more nets with multiple
drivers, it would have to be iterated to fixpoint. Moreover, even if
this was done, it would not be sufficient because `opt_clean -purge`
does not currently remove wires with the `\init` attribute (and it
is not desirable to remove such wires, since they correspond to
registers and may be useful for debugging).
The proper solution is to consider the condition in which a wire
may be localized. Specifically, if there are no feedback arcs through
this wire, and no part of the wire is driven by an output of a sync
cell, then the wire holds no state and is localizable.
After this commit, the original condition for not localizing a wire
is replaced by a check for any sync cell driving it. This makes it
unnecessary to run `splitnets -driver` in the majority of cases
to get a design with no buffered wires, and -O5 no longer includes
that pass. As a result, Minerva SRAM SoC no longer has any buffered
wires, and runs ~27% faster.
In addition, this commit prepares the flow graph for introduction
of sync outputs of black boxes.
Co-authored-by: Jean-François Nguyen <jf@lambdaconcept.com>
whitequark [Mon, 20 Apr 2020 16:44:51 +0000 (16:44 +0000)]
cxxrtl: detect buffered comb wires, not just feedback wires.
Any buffered combinatorial wires (including, as a subset, feedback
wires) will prevent the design from always converging in one delta
cycle. Before this commit, only feedback wires were detected. After
this commit, any buffered combinatorial wires, including feedback
wires, are detected.
Co-authored-by: Jean-François Nguyen <jf@lambdaconcept.com>
Claire Wolf [Tue, 21 Apr 2020 12:23:24 +0000 (14:23 +0200)]
Add '=' selection pattern prefix for non-blackbox only patterns
Signed-off-by: Claire Wolf <claire@symbioticeda.com>
Claire Wolf [Tue, 21 Apr 2020 10:35:25 +0000 (12:35 +0200)]
Improve net priorities in EDIF back-end
Signed-off-by: Claire Wolf <claire@symbioticeda.com>
David Shah [Tue, 21 Apr 2020 07:58:52 +0000 (08:58 +0100)]
sim: Fix handling of constant-connected cell inputs at startup
Signed-off-by: David Shah <dave@ds0.me>
whitequark [Tue, 21 Apr 2020 01:43:20 +0000 (01:43 +0000)]
Merge pull request #1961 from whitequark/paramod-original-name
ast, rpc: record original name of $paramod\* as \hdlname attribute
Eddie Hung [Mon, 20 Apr 2020 22:42:29 +0000 (15:42 -0700)]
tests: remove write_ilang
Eddie Hung [Mon, 20 Apr 2020 22:34:31 +0000 (15:34 -0700)]
Merge pull request #1975 from dh73/claire/bitselwrite
Adding tests to Claire/bitselwrite branch
Eddie Hung [Mon, 20 Apr 2020 18:58:23 +0000 (11:58 -0700)]
Remove '-ignore_unknown_cells' option from 'sat'
Eddie Hung [Mon, 20 Apr 2020 18:54:10 +0000 (11:54 -0700)]
Simplify test case script
Eddie Hung [Mon, 20 Apr 2020 18:53:48 +0000 (11:53 -0700)]
Remove ununsed files
Eddie Hung [Mon, 20 Apr 2020 18:39:13 +0000 (11:39 -0700)]
Merge pull request #1972 from YosysHQ/eddie/bug1970
abc9_ops: -prep_lut to be more robust
Eddie Hung [Mon, 20 Apr 2020 18:26:11 +0000 (11:26 -0700)]
abc9: tolerate ABC nonzero exit code if output.aig; write before &mfs
Re-enable mfs for xilinx/ecp5 speculatively -- if it fails, use pre-mfs
result
diego [Mon, 20 Apr 2020 17:45:35 +0000 (12:45 -0500)]
Modifications of tests as per Eddie's request
Eddie Hung [Mon, 20 Apr 2020 17:30:10 +0000 (10:30 -0700)]
xilinx/ecp5: disable abc9's "&mfs" optimisation
Can sometimes fire an assertion, e.g. #1962
Eddie Hung [Mon, 20 Apr 2020 16:39:35 +0000 (09:39 -0700)]
abc9: -prep_lut to be more robust
Eddie Hung [Mon, 20 Apr 2020 16:38:29 +0000 (09:38 -0700)]
abc9: add testcase reduced from #1970
Claire Wolf [Mon, 20 Apr 2020 14:00:37 +0000 (16:00 +0200)]
Ignore conflicting keep attributes, unless asked not to. Fixes #1733
Signed-off-by: Claire Wolf <claire@symbioticeda.com>
Claire Wolf [Mon, 20 Apr 2020 12:51:40 +0000 (14:51 +0200)]
Merge pull request #1964 from YosysHQ/claire/sformatf
Extend support for format strings in Verilog front-end
Alberto Gonzalez [Mon, 20 Apr 2020 05:37:10 +0000 (05:37 +0000)]
Add rvalue-reference-accepting `entry_t` constructor for `pool`.
Alberto Gonzalez [Mon, 20 Apr 2020 02:16:55 +0000 (02:16 +0000)]
In `pool`, construct `entry_t`s in-place and add an rvalue-accepting-and-forwarding `insert()` method.
whitequark [Sun, 19 Apr 2020 19:59:42 +0000 (19:59 +0000)]
Merge pull request #1967 from whitequark/cxxrtl-blackbox-attributes
cxxrtl: provide attributes to black box factories, too
whitequark [Sun, 19 Apr 2020 16:22:02 +0000 (16:22 +0000)]
cxxrtl: provide attributes to black box factories, too.
Both parameters and attributes are necessary because the parameters
have to be the same between every instantiation of the cell, but
attributes may well vary. For example, for an UART PHY, the type
of the PHY (tty, pty, socket) would be a parameter, but configuration
of the implementation specified by the type (socket address) would
be an attribute.
Claire Wolf [Sat, 18 Apr 2020 12:08:51 +0000 (14:08 +0200)]
Extend support for format strings in Verilog front-end
Signed-off-by: Claire Wolf <claire@symbioticeda.com>
whitequark [Sat, 18 Apr 2020 09:21:14 +0000 (09:21 +0000)]
Merge pull request #1963 from whitequark/cxxrtl-blackboxes
cxxrtl: add support for simple and templated C++ black boxes
whitequark [Sat, 18 Apr 2020 08:04:57 +0000 (08:04 +0000)]
cxxrtl: add templated black box support.
whitequark [Sat, 18 Apr 2020 02:14:20 +0000 (02:14 +0000)]
cxxrtl: make eval() and commit() inline in blackboxes.
This change is a preparation for template blackboxes. It has no
effect on current generated code.
whitequark [Fri, 17 Apr 2020 01:41:08 +0000 (01:41 +0000)]
cxxrtl: add simple black box support.
This commit adds support for replacing RTLIL modules with CXXRTL
black boxes. Black box port widths may not depend on the parameters
with which it is instantiated (yet); the parameters may only be used
to change the behavior of the black box.
whitequark [Wed, 15 Apr 2020 17:39:14 +0000 (17:39 +0000)]
cxxrtl: use ID::X instead of ID(X). NFC.
whitequark [Sat, 18 Apr 2020 03:43:15 +0000 (03:43 +0000)]
ast, rpc: record original name of $paramod\* as \hdlname attribute.
The $paramod name mangling is not invertible (the \ character, which
separates the module name from the parameters, is valid in the module
name itself), which does not stop people from trying to invert it.
This commit makes it easy to invert the name mangling by storing
the original name explicitly, and fixes the firrtl backend to use
the newly introduced attribute.
whitequark [Fri, 17 Apr 2020 21:36:59 +0000 (21:36 +0000)]
Merge pull request #1955 from whitequark/cxxrtl-sync_always
cxxrtl: correctly handle `sync always` rules
whitequark [Fri, 17 Apr 2020 18:57:00 +0000 (18:57 +0000)]
Merge pull request #1952 from boqwxp/add_edge_location
Verilog frontend: add source location in more parser rules
diego [Fri, 17 Apr 2020 15:15:22 +0000 (10:15 -0500)]
Wrong fixed value
whitequark [Fri, 17 Apr 2020 09:43:13 +0000 (09:43 +0000)]
cxxrtl: correctly handle `sync always` rules.
Fixes #1948.
whitequark [Fri, 17 Apr 2020 09:07:49 +0000 (09:07 +0000)]
Merge pull request #1954 from YosysHQ/dave/fix-stdout-conflict
qbfsat: Fix illegal use of 'stdout' identifier
whitequark [Fri, 17 Apr 2020 09:06:40 +0000 (09:06 +0000)]
Merge pull request #1951 from whitequark/rtlil-string_attribute
rtlil: add AttrObject::{get,set}_string_attribute, AttrObject::has_attribute
David Shah [Fri, 17 Apr 2020 07:42:39 +0000 (08:42 +0100)]
qbfsat: Fix illegal use of 'stdout' identifier
Signed-off-by: David Shah <dave@ds0.me>
Alberto Gonzalez [Fri, 17 Apr 2020 06:23:03 +0000 (06:23 +0000)]
Set Verilog source location for explicit blocks (`begin` ... `end`).
Alberto Gonzalez [Fri, 17 Apr 2020 06:16:59 +0000 (06:16 +0000)]
Add Verilog source location information to `AST_POSEDGE` and `AST_NEGEDGE` nodes.
whitequark [Fri, 17 Apr 2020 02:27:13 +0000 (02:27 +0000)]
Merge pull request #1898 from boqwxp/locations
Verilog frontend: add location information to parsed constants
whitequark [Fri, 17 Apr 2020 02:25:18 +0000 (02:25 +0000)]
Merge pull request #1864 from boqwxp/cleanup_techmap_abc
Clean up pseudo-private member usage and simplify `passes/techmap/abc.cc`
whitequark [Fri, 17 Apr 2020 02:21:23 +0000 (02:21 +0000)]
Merge pull request #1888 from boqwxp/cleanup_scatter
Clean up `passes/cmds/scatter.cc`.
whitequark [Fri, 17 Apr 2020 02:20:54 +0000 (02:20 +0000)]
Merge pull request #1882 from boqwxp/cleanup_rename
Clean up pseudo-private member usage in `passes/cmds/rename.cc`.
whitequark [Thu, 16 Apr 2020 22:09:25 +0000 (22:09 +0000)]
Merge pull request #1929 from YosysHQ/eddie/select_unset
select: add select -unset option
whitequark [Thu, 16 Apr 2020 21:49:49 +0000 (21:49 +0000)]
rtlil: add AttrObject::has_attribute.